Characteristics of electroencephalographic changes induced by different hypnotics in elderly patients: a narrative review

Abstract

Aging is associated with widespread structural and functional changes in the brain including reduced neural plasticity, slower information processing, and impaired network integration. These age-related alterations influence the brain’s response to anesthetic agents, particularly electroencephalography (EEG) activity. This narrative review summarizes the characteristic EEG features induced by commonly used hypnotic agents such as propofol, inhaled anesthetics, dexmedetomidine, ketamine, and remimazolam in elderly patients and examines how aging modulates these responses. With increasing age, EEG power shows a global decline, most prominently in the alpha frequency band (8–13 Hz), reflecting reduced thalamocortical and cortical activity. Peak alpha frequency slows progressively with age, and background EEG also often exhibits characteristic slowing, both of which are associated with cognitive decline. In addition, EEG reactivity to external stimuli diminishes, and integrative brain activity, representing coordinated processing across cortical regions, is reduced in older adults. Frontoparietal feedback connectivity, essential for conscious perception and information integration, is particularly weak in the elderly. These changes are further exacerbated under anesthesia, as general anesthetics disrupt top-down connectivity and reduce network integration. Graph-theoretical EEG analyses reveal age-related reductions in global efficiency, modularity, and small-world properties, which are signatures of a less efficient, more random, and fragmented brain network. Understanding these age-specific EEG alterations can improve intraoperative monitoring, anesthetic titration, and development of age-tailored EEG-guided strategies. Future research should aim to validate EEG biomarkers that reliably reflect anesthetic depth and brain health in elderly populations, thereby fostering safer anesthesia care in the aging population.

Introduction

Words like “youth” and “novelty” stir excitement and vitality in our hearts. However, eternal youth is an illusion. Qin Shi Huang, the first emperor of unified China, is alleged to have devoted much of his life to seeking an elixir of immortality. However, aging is an inevitable part of life, and returning to nature is a universal truth. In South Korea, clinical encounters with patients aged 65 years and older have become increasingly common as the country has not only become an aged society but, by 2025, has transitioned into a super-aged society [1]. Elderly patients often exhibit delayed responses to anesthetic agents compared to younger adults, and typically require lower doses to achieve the same clinical effect [2]. Therefore, precise dose titration is essential. Moreover, age affects many current processed electroencephalography (EEG) indices, raising concerns regarding the risk of anesthetic overdose in this population, as demonstrated in previous studies [3,4].

As titration guided by processed EEG indices is becoming more widely adopted in general anesthesia practice, a deep understanding of EEG characteristics in elderly patients is essential for proper anesthetic management. This narrative review aims to explore the neurophysiological changes associated with aging, provide a conceptual overview of EEG measurements and analysis from a clinical perspective, and examine the EEG patterns of various commonly used anesthetics, with a particular focus on their features in elderly patients. In particular, we emphasize how aging and anesthesia affect large-scale brain connectivity, including both pairwise and network-level interactions, and how these changes are reflected in EEG-based functional and effective connectivity measures as well as graph-theoretical network properties such as efficiency and modularity.

Aging brain

Aging refers to the gradual decline in biological functions associated with increasing age. It is typically characterized by reduced resilience to stress, impaired homeostasis, and increased susceptibility to disease [5]. The rate of aging may vary depending on individual genetic and environmental factors, indicating that biological age does not always correspond precisely to chronological age. Aging of the nervous system, including the brain, is a natural and irreversible process that encompasses molecular, cellular, and structural changes, ultimately leading to functional decline [6].

The volume and weight of the brain decrease by approximately 5% per decade after the age of 40 years, and accelerate after the age of 70 years [7]. Volume loss is relatively diffuse and uniform across the white matter [7]. However, among the grey matter structures, the frontal and parietal cortices and the striatum are more affected than the temporal cortex, cerebellar vermis, and hippocampus. The occipital cortex is the least affected [8]. The regional susceptibility of brain structures to age-related changes are shown in Table 1. Findings suggesting that the frontal cortex is most affected and the occipital cortex is least affected aligns with the cognitive changes observed with aging [9]. In particular, reductions in hippocampal volume and neurogenesis have been associated with aging [9,10]. The hippocampus plays a central role in organizing, storing, and retrieving memory, and converting short-term memory into long-term memory [11]. Several studies have reported that aging-induced structural and functional changes in the hippocampus are associated with cognitive decline [12,13].

Table 1.

Regional Susceptibility of Brain Structures to Age-related Changes

Region	Degree of decline
	High	Moderate	Mild	Least
Frontal cortex	O
Parietal cortex	O
Striatum		O
Temporal cortex			O
Cerebellar vermis			O
Hippocampus			O
Occipital cortex				O

The frontal and parietal cortices as well as the striatum exhibited the most pronounced atrophy, whereas the temporal cortex, cerebellar vermis, and hippocampus showed milder reduction. The occipital cortex was the least affected. Notably, the hippocampus exhibited both volume reduction and decreased neurogenesis, both of which are associated with cognitive decline in older adults.

Additionally, aging is accompanied by a reduction in neurotransmitters, such as dopamine and acetylcholine [14,15]. However, serotonin levels appear to remain relatively stable during the normal aging process [14]. In a study assessing regional cerebral blood flow and oxygen consumption in healthy individuals aged 19–76 years, a linear decline in mean cerebral blood flow in the gray matter was observed [16]. When the cohort was divided based on a cut-off age of 50 years, significant reductions in the cerebral blood flow and cerebral metabolic rate of oxygen were noted in the gray matter of the older group [16]. In contrast, white-matter cerebral perfusion and metabolism were relatively preserved with age [16].

Overview of EEG analysis for clinicians

EEG records wave-like bioelectrical signals generated when neurons in the brain communicate with each other [17]. The biosignals detected by EEG primarily reflect the postsynaptic potentials of pyramidal neurons in both the neocortex and allocortex [17]. EEG is typically noninvasive, with electrodes placed along the scalp according to the International 10–20 system or its variations [18].

Spectral analysis

Because EEG patterns change with the level of consciousness, EEG is widely used as an indicator of hypnotic depth. However, detecting subtle changes in hypnotic depth by visually interpreting raw EEG data in real-time can be challenging. Therefore, frequency domain analyses are typically performed [19]. This approach requires an understanding of sampling (or digitization), which involves converting analog signals into a digital form. Digital signal analysis systems used for EEG signal processing must collect EEG data at regular intervals. The recorded EEG signals can then be decomposed into sinusoids, which are defined by three basic parameters: amplitude, frequency, and phase angle [20]. Fig. 1 illustrates how a sinusoid changes with variations in phase angle, frequency, and amplitude. For EEG, the amplitude is most commonly expressed in microvolts (µV). The phase angle (θ) represents the temporal offset of the sinusoid relative to time zero and is expressed as a fraction of a full cycle. It is typically measured in degrees (°), ranging from 0° to 360°, with 360° indicating a complete cycle. In practice, the phase angles are often expressed in radians. A sinusoid can be visualized as a vector of magnitude and direction. The magnitude (or length) of the vector corresponds to the peak amplitude of the sine wave. The direction of the vector is the phase angle θ, measured counterclockwise from the horizontal axis. This vector representation can be expressed visually by projecting the components onto real and imaginary axes. Regardless of its complexity, any continuous EEG signal can be expressed as the sum of sinusoidal components, known as a Fourier series. Transforming a signal into its Fourier series allows for the individual analysis of each sinusoidal component, which forms the foundation of quantitative EEG analysis [20].

Fig. 1.

Graphical representation of the three fundamental parameters of a sinusoidal EEG signal: amplitude, frequency, and phase angle. The Y-axis represents amplitude in µV. (A) Signals with different amplitudes, (B) Signals with different frequencies, and (C) Signals with different phase angles. Each parameter affects the waveform’s appearance and forms the basis for frequency-domain EEG analysis. EEG: electroencephalography.

EEG signals are commonly described using frequency bands such as delta (δ, 1–4 Hz), theta (θ, 4–8 Hz), alpha (α, 8–13 Hz), beta (β, 13–30 Hz), and gamma (γ, 30–80 Hz) [21,22]. Frequencies < 1 Hz are referred to as slow waves [21]. The process of transforming a time-domain signal into its frequency-domain representation in terms of sinusoidal amplitudes and frequencies is known as Fourier transform. This transformation converts a time series signal x (t), where the amplitude or power is a function of time, into a frequency series signal X (f), where the amplitude or power is a function of the frequency. To accurately capture a sinusoidal component, the sampling frequency (f_s) must be at least twice the frequency of the signal (f), in accordance with the Nyquist theorem (f_s ≥ 2×f). If the sampling frequency is too low and fewer than two samples are obtained per cycle, aliasing may occur. Aliasing is the distortion that occurs when the higher-frequency components of the original analog signal are incorrectly represented as lower-frequency components in the digitized signal (Fig. 2). To prevent aliasing in Fourier analysis, the maximum frequency of interest must be less than half the sampling frequency, known as the Nyquist frequency (f_s/2) [23]. Oversampling beyond the Nyquist rate substantially increases the data size, computational burden, and susceptibility to high-frequency noise without improving the signal fidelity. Thus, setting an appropriate sampling frequency during data acquisition is critical to preserve the desired frequency range for analysis.

Fig. 2.

Demonstration of aliasing due to insufficient sampling and correct signal reconstruction with adequate sampling. The top panel shows a true sinusoidal signal at 30 Hz (black line) with sampled points at 10 Hz (red crosses) and 60 Hz (blue circles). When the 30-Hz signal is sampled at 10 Hz, the signal is erroneously perceived as a 5-Hz waveform due to aliasing, as shown in the middle panel (red line). In contrast, sampling at 60 Hz correctly reconstructs the original 30-Hz signal without aliasing, as demonstrated in the bottom panel (blue line). These examples illustrate the importance of satisfying the Nyquist criterion for accurate signal representation.

Power spectral analysis quantifies the distribution of power across different frequencies while ignoring the phase angle information. The underlying signal is assumed to be generated through a linear system. As a result, this method overlooks phase coupling, which reflects the interactions between different signal components. Phase coupling is a common feature of signals produced by nonlinear systems such as the central nervous system. Bispectral analysis is another method used to quantify the degree of phase coupling between EEG signal components [24]. This can be viewed as a two-dimensional Fourier transform. In nonlinear systems, sinusoidal input signals often produce dependent sinusoidal components in their output. When the sinusoidal output components arise from multiplication (rather than addition or subtraction) of the input components, the resulting components are called intermodulation products. Components that do not result from such interactions are referred to as fundamental components. The presence of intermodulation products indicates that phase coupling has occurred [24]. The presence of phase coupling cannot be determined from a single epoch. Multiple epochs must be examined to determine whether the phase relationships between the frequency pairs are random or systematic. This is achieved by computing the triple product for each epoch and averaging the results. The magnitude of this averaged triple product represents the bispectrum [24]. A normalized measure called bicoherence is used to assess the strength of the phase coupling. Bicoherence standardizes the bispectrum on a scale from 0% to 100% [25]. If all the sinusoidal components of the EEG signal are fundamental (i.e., not phase-coupled), then the bispectrum and bicoherence will both be zero. Conversely, if all components are perfectly phase-coupled, the bispectrum will equal the real part of the triple product, and the bicoherence will reach 100%. In the intermediate cases where partial phase coupling exists, the bicoherence values will fall somewhere between 0% and 100%.

Connectivity analysis

Over the past two decades, EEG connectivity analyses have provided key insights into how anesthetic drugs alter the brain’s network dynamics to induce and reverse unconsciousness. Unlike spectral power analyses, which capture local oscillatory changes, connectivity measures reveal how communication across cortical and subcortical regions is disrupted and restored during anesthesia [26–28]. Connectivity measures are broadly categorized into pairwise connectivity (e.g., functional and effective connectivity) and graph-theoretical network connectivity involving multiple EEG channels. Functional connectivity captures the statistical relationships (e.g., correlation or synchronization) between EEG signals from different brain regions, reflecting temporally coordinated but non-causal interactions [29,30]. By contrast, effective connectivity quantifies directed causal influences between EEG signals using methods such as Granger causality or transfer entropy [29,31,32].

Graph-theoretical network analysis, grounded in graph theory, extends beyond pairwise measures by modeling the brain as a network of nodes (e.g., EEG channels, functional clusters, or brain regions) interconnected by edges (e.g., functional or structural connections) to characterize the global and local organization of information flow [27,28]. Additionally, network dynamics extend the spatial domain of graph-theoretical network analysis into the temporal domain, capturing how patterns of neural activity evolve over time within the brain’s structural or functional network. These dynamics can be quantified using measures of spatiotemporal complexity that capture the richness and diversity of the brain network’s dynamic repertoire [33,34]. Among the various complexity metrics, including different forms of entropy, the perturbational complexity index (PCI) has emerged as a powerful tool [35]. The PCI is widely recognized for reliably quantifying the levels of consciousness across diverse conditions, including sleep; anesthesia; and disorders of consciousness, such as unresponsive wakefulness syndrome, minimally conscious states, locked-in syndrome, and coma [36,37]. Simultaneously, the criticality framework, which originates from thermodynamics, offers a unified perspective on how consciousness emerges and fades across diverse conditions. “Criticality” refers to the balanced state in which the brain operates between order and disorder, stability and instability, or segregation and integration. At this critical point, the brain maintains optimal information processing, adaptability, and energy efficiency along with maximal complexity (e.g., PCI) [38–42].

General characteristics of EEG changes during general anesthesia

Patients are typically alert before administration of hypnotic agents. When instructed to remain still with their eyes closed, alpha band activity becomes predominant on the EEG [43]. The administration of a small dose of a hypnotic agent, such as propofol, induces a state of sedation during which the patient appears calm, closes their eyes, and can be easily aroused by external stimuli [44]. As the dose of the hypnotic agent gradually increases, patients may enter a paradoxical excitation phase characterized by purposeless movements or incoherent speech [45]. This stage is associated with increased beta-band activity and is particularly associated with propofol [46]. The EEG patterns observed during the progressive depths of general anesthesia are shown in Fig. 3. Once consciousness is lost and the hypnotic depth deepens, four characteristic EEG patterns can be observed [47]. In the first stage, which represents light general anesthesia, beta band activity decreases, whereas alpha and delta band activity increases [48]. As the hypnotic state deepens further, a phenomenon known as “anteriorization” emerges, wherein the decrease in beta activity and the increase in alpha and delta activity becomes more prominent in the anterior regions of the brain than in the posterior regions [48,49]. The EEG signal during this stage resembles that observed during non-rapid eye movement (NREM) or slow-wave sleep [43]. As the depth of anesthesia progresses to the third stage, periods of flat (isoelectric) EEG activity begin to appear between bursts of alpha- and beta-band activity, a phenomenon commonly referred to as burst suppression [50]. As the depth of anesthesia increases further, the intervals between bursts become longer, and the amplitude of alpha and beta activity within the bursts diminishes [43].

Fig. 3.

Electroencephalography (EEG) patterns observed during progressive depths of general anesthesia. Panels A through E illustrate simulated EEG waveforms corresponding to characteristic stages of anesthesia. (A) Awake state with dominant alpha activity, (B) Sedation with mixed alpha and low-amplitude beta activity, (C) Paradoxical excitation marked by increased beta activity, (D) Light general anesthesia showing combined alpha and delta activity, and (E) Deep anesthesia exhibiting a burst suppression pattern with alternating periods of high-amplitude activity and isoelectric suppression. These patterns reflect dynamic changes in cortical activity and serve as neurophysiological markers of hypnotic depth.

EEG connectivity analyses have potentially revealed a common mechanism underlying diverse anesthetics despite their distinct molecular targets and neurophysiologic actions [51–55]. In an effort to find a unified mechanism for anesthetic-induced loss and recovery of consciousness, ketamine has long been an exception. Unlike GABAergic agents such as propofol or sevoflurane, ketamine paradoxically increases cortical activation, high-frequency EEG activity, and signal complexity, which are features usually associated with wakefulness, yet patients remain unresponsive. EEG connectivity studies have revealed striking convergence across different anesthetics, showing a consistent breakdown of frontal-to-parietal (top-down or feedback) connectivity (Fig. 4) [52–57]. This pathway may support the integration of information across sensory modalities, which is a core requirement for conscious awareness. During anesthesia, frontal-to-parietal (top-down) connectivity collapses, whereas parietal-to-frontal (bottom-up) connectivity remains relatively preserved [58–61]. In other words, the brain continues to receive sensory inputs through bottom-up pathways but can no longer integrate them into coherent percepts through top-down processing.

Fig. 4.

Distinct network organizations of healthy, old, and anesthetized brains. (A) Healthy brain: complex hierarchical network with efficient communication between distant regions, especially front-to-back (feedback) connectivity supporting awareness. (B) Old brain: connections are weaker, more random, and less efficient. Long-range front-to-back connectivity fades, and communication becomes slower and uneven. (C) Anesthetized brain: major communication connectivity, particularly from frontal to back region, are disrupted. The network is fragmented into small, isolated clusters, reflecting loss of global coordination and consciousness.

Measures such as transfer entropy and the directed phase lag index show that this feedback loss reverses upon recovery, underscoring its critical role in both the loss and restoration of consciousness [53–55]. Thus, although anesthetics act on distinct molecular targets, they converge at the network level by fragmenting large-scale cortical communication. While some studies have reported counterexamples such as altered parietal-to-frontal rather than frontal-to-parietal connectivity [51,62], many follow-up studies consistently support the preferential disruption of frontal-to-back (parietal or posterior; top-down) connectivity across different anesthetics and species, including humans, macaques, rodents, and fruit flies [53,63–65].

A similar loss of integrative connectivity has also been observed in aging and delirium, where reduced interregional coordination contributes to cognitive slowing and confusion [66,67], suggesting a potential network mechanism shared by both pharmacological and age-related impairments in brain function. Under anesthesia, graph metrics such as decreased global efficiency [68,69], reduced small-worldness (a network configuration that balances local clustering with global communication) [70–72], and fragmentation of long-range connections [73–75] indicate a loss of large-scale brain integration that is essential for the emergence of consciousness. This breakdown of brain communication under anesthesia is similar to that which occurs in a major air traffic system during a winter storm. When heavy snow hits large hub airports such as O’Hare in Chicago or John F. Kennedy in New York during the busy holiday season, long-distance flights are delayed or canceled, and the entire international flight network becomes fragmented. Local flights may still take off, but global traffic coordination collapses, slows down, or even stops air travel. In this analogy, snow represents GABA-based anesthetics such as propofol or sevoflurane. These drugs inhibit brain activity and block communication between regions, preferentially targeting hub regions with intensive connections, similar to snow shutting down hub airports and disrupting normal air traffic flow. In contrast, ketamine anesthesia works differently. This can be compared to a city during a large festival or street carnival. The streets are full of tourists and noise, and people are excited; however, the traffic becomes chaotic and disorganized. Similarly, ketamine over-activates neurons, creating excessive random communication. Because activity levels are excessively high, the brain loses coordination, and normal information flow breaks down.

These analogies illustrate how anesthetic drugs, despite acting on different molecular targets (GABAergic or NMDA receptors), converge on a common network-level effect, disrupting long-range communication pathways, fragmenting brain connectivity, and preventing coordinated information exchange, ultimately resulting in unconsciousness. When physiological, pharmacological, or pathological perturbations push the brain away from this balance toward either excessive integration or segregation, consciousness begins to fade. In an overly integrated state (i.e., supercritical), such as during an epileptic seizure, global neuronal activities in the brain are highly synchronized, eliminating the diversity of activity patterns required for meaningful information processing. Conversely, in an overly segregated state, as seen in a coma or deep anesthesia, brain activity becomes fragmented and uncoordinated, preventing the large-scale integration required for awareness. Thus, the degree to which the brain deviates from critical balance has been proposed as a metric to assess the level of consciousness [76–79]. Furthermore, a recent study showed that the brain’s type of phase transition (first- or second-order) near a critical balance, measured in the conscious resting state before anesthesia, predicts whether an individual will lose or regain consciousness rapidly or gradually [80].

In summary, EEG connectivity studies have shown that anesthesia-induced unconsciousness arises from the disruption of large-scale brain integration, particularly the breakdown of frontal-to-parietal (top-down) connectivity, which supports conscious perception. Graph-theoretical analyses reveal a loss of global efficiency, small-world organization, and long-range coordination, analogous to an air-traffic system paralyzed by a snowstorm for GABAergic agents, or chaos from overcrowding during a festival for ketamine. Despite their different molecular actions, both GABAergic and NMDA-antagonist anesthetics fragment brain networks and push them away from an optimally balanced state between segregation and integration, which is essential for consciousness.

Distinctive EEG signatures induced by individual hypnotic agents

A summary of the EEG features associated with each hypnotic agent is provided in Table 2.

Table 2.

Electroencephalographic Features Associated with Each Hypnotic Agent

Anesthetic agent	Frequency features	Typical pattern
Propofol	↑ alpha (8–13 Hz), ↑ delta (1–4 Hz)	Frontally dominant alpha oscillations with slow-wave enhancement
Sevoflurane	• MAC < 1: ↑ alpha and delta	Broadband enhancement across delta, theta, and alpha; unique spectral pattern on DSA. Desflurane and isoflurane show generally similar patterns
	• MAC > 1: ↑ theta (4–8 Hz), alpha, and delta
Dexmedetomidine	Spindle-like alpha, ↑ slow-delta	Sleep-like EEG resembling stage 2 NREM; prominent spindles and slow waves
Ketamine	↑ gamma (> 30 Hz), ↓ alpha	Irregular, noisy pattern; separation of slow oscillations and high-frequency activity
Remimazolam	↑ beta (13–30 Hz), ↑ alpha	Beta activity during sedation; propofol-like pattern after loss of consciousness

MAC: minimum alveolar concentration, DSA: density spectral array, EEG: electroencephalography, NREM: non-rapid eye movement.

Propofol

The EEG patterns associated with propofol administration vary depending on the rate of drug delivery. Although EEG dynamics differ between bolus injections and slow continuous infusions, a common feature observed during anesthetic induction is the transition from high-frequency gamma and beta oscillations, typically observed in the awake state, to prominent slow delta oscillations [21]. The amplitude of slow delta oscillations has been reported to be 5–20 times greater than those of gamma and beta oscillations observed during wakefulness [81]. The co-occurrence of alpha oscillations and slow delta activity is a characteristic feature seen at the point that consciousness is lost (Fig. 5A) [21,82]. Notably, these alpha oscillations disappear from the occipital region and become prominent in the frontal region, a phenomenon referred to as anteriorization (Fig. 5B) [82]. Upon discontinuation of propofol infusion, patients gradually transition to a wakeful state. During this emergence process, slow delta and alpha oscillations progressively dissipate and are replaced by lower-amplitude, higher-frequency beta and gamma oscillations [21]. The reversal of anteriorization has also been observed during emergence [82].

Fig. 5.

Spectral and topographic characteristics of frontal electroencephalography (EEG) during propofol-induced unconsciousness. (A) Time–frequency spectrogram of frontal EEG activity recorded from a 32-year-old volunteer during continuous propofol infusion. The spectrogram depicts EEG power across frequencies from 0 to 40 Hz over a 10-min window. Prominent slow-delta (0.1–4 Hz) and alpha (8–12 Hz) oscillations are observed during this period, consistent with the characteristic EEG signatures of propofol-induced unconsciousness. Power is expressed in decibels (dB), as indicated by the color bar. (B) Scalp topographic maps illustrating anteriorization of alpha-band (8–12 Hz) power following propofol administration. The left panel represents the baseline awake state prior to infusion, while the right panel shows the distribution after loss of consciousness. A marked anterior shift of alpha power is evident after the onset of unconsciousness, reflecting the typical anteriorization pattern associated with propofol anesthesia. Power is expressed in dB, as indicated by the color bar.

Inhaled anesthetics

EEG changes induced by inhalational anesthetics, such as sevoflurane, desflurane, and isoflurane, exhibit similar patterns [21]. During sevoflurane administration at concentrations below the minimum alveolar concentration (MAC), prominent alpha and delta oscillations closely resembling the EEG patterns are observed with propofol [21]. When the concentration of sevoflurane exceeds the MAC, theta oscillations emerge, producing a distinct pattern characterized by broad activation from slow delta to alpha bands, as shown in the density spectral array (Fig. 6) [21]. Because end-tidal sevoflurane concentrations are typically maintained above the MAC during general anesthesia, this characteristic pattern is commonly observed. As the sevoflurane concentration decreases, the theta oscillations are the first to disappear. During emergence from anesthesia, both the alpha and delta oscillations gradually diminish [21].

Fig. 6.

Spectrogram of frontal electroencephalography (EEG) activity during maintenance of anesthesia with sevoflurane. The spectrogram displays EEG power across frequencies (0–30 Hz) recorded while maintaining an end-tidal sevoflurane concentration of 2.5%. Distinct oscillatory patterns are evident, including slow (< 1 Hz), delta (1–4 Hz), theta (4–8 Hz), and alpha (8–12 Hz) bands. These frequency components represent characteristic EEG signatures observed during steady-state sevoflurane anesthesia. Power is expressed in decibels, as indicated by the color scale.

Dexmedetomidine

One of the most prominent EEG features of dexmedetomidine sedation is the presence of spindles, which are brief bursts of 9–15 Hz oscillations lasting 1–2 s, accompanied by slow-delta oscillations (Fig. 7) [21]. These spindles resemble the alpha oscillations observed during propofol anesthesia, although they have a significantly lower power [81]. The spindles are similar to those characterized by stage II NREM sleep [83]. A large-scale validation study using machine learning has also reported that the deep sedation state induced by dexmedetomidine resembles stage III NREM sleep [84]. As the concentration of dexmedetomidine increases, the spindles diminish and the amplitude of delta oscillations increases [21]. As the patient approaches wakefulness, the EEG alpha and delta powers progressively decrease, whereas the beta power increases [85].

Fig. 7.

Spindle activity observed on electroencephalography (EEG) during dexmedetomidine sedation. Representative EEG trace showing spindle activity (red boxes) during sedation with dexmedetomidine. Spindles are defined as brief bursts of 9–15 Hz oscillations lasting 1–2 s and are a hallmark EEG feature of dexmedetomidine.

Ketamine

At sub-anesthetic doses, ketamine induces oscillatory activity in the 25–32 Hz range, commonly referred to as gamma oscillations [21]. At anesthetic doses, ketamine-induced unconsciousness is characterized by the presence of gamma bursts, which are alternating patterns of synchronized gamma and slow-delta oscillations [86]. In addition, a reduction in alpha oscillation power is typically observed, which has been associated with ketamine-induced dissociation [87].

Remimazolam

Remimazolam is a relatively new hypnotic agent that was initially approved for general anesthesia in Japan and South Korea and more recently approved in Europe in 2024. Consequently, studies investigating the EEG characteristics of remimazolam are limited compared with those investigating other anesthetic agents. Given that remimazolam shares pharmacological properties with midazolam, the EEG features observed with midazolam may similarly be observed with remimazolam. After loss of consciousness following a single bolus of midazolam, prominent beta band activity (13–30 Hz) has been observed, followed by the emergence of alpha band activity [88]. This pattern has similarly been reported in studies using remimazolam for sedation [89]. However, in a study in which remimazolam was continuously infused at 6 mg/kg/h to induce general anesthesia, beta oscillations were not clearly observed, whereas delta and alpha oscillations resembling those of propofol were observed [90]. Anteriorization, characterized by the emergence of alpha oscillations in the frontal region following loss of consciousness, has also been observed [90].

Age-related EEG changes

Age-related neurophysiological alterations are commonly observed in the EEG patterns of older adults. A general reduction in the electrical activity of the brain is observed with aging, leading to a global decrease in EEG power [91]. This reduction is particularly evident in the alpha frequency band (8–13 Hz), whereas a relative increase in delta (1–4 Hz) and theta (4–8 Hz) powers has been observed. Consequently, slowing of background EEG is frequently observed in elderly individuals [92]. Age-related slowing of peak alpha frequency is well documented [93]. Reductions in alpha power or shifts in peak alpha frequency have been associated with cognitive decline and proposed as potential biomarkers for the early detection of neurodegenerative diseases [94,95]. Furthermore, intraoperative slowing of peak alpha frequency has been independently linked to postoperative delirium in patients undergoing cardiac surgery [96]. A separate study also reported that a lower intraoperative alpha power was associated with a higher incidence of postoperative delirium [97]. Age-related changes in frequency bands vary across different brain regions. Specifically, cortical alpha activity decreases with age in the parietal, occipital, and temporal regions [98]. In contrast, delta and theta powers have been reported to decrease in the occipital cortex [99]. Beta activity appears to increase in the insular cortex, whereas gamma activity tends to increase in the frontoparietal regions [99]. Although some studies have reported conflicting findings regarding age-related EEG changes [100,101], which may be attributed to differences in subject characteristics and analytical methodologies, aging is increasingly understood to be associated with region-specific alterations in EEG frequency bands. Additionally, EEG reactivity to external stimuli is diminished in older adults and integrative brain activity, defined as the coordinated processing of information across multiple brain regions, appears to decrease with aging [102]. These findings are summarized in Table 3.

Table 3.

Summary of Age-related EEG Changes

EEG feature	Age-related change	Regional specificity	Notes
Global EEG power	Decreased	Whole brain	General slowing of brain electrical activity
Alpha power (8–13 Hz)	Decreased	Parietal, occipital, temporal	Most consistently reduced band with aging
Peak alpha frequency	Slowing	General	Well-established age-related slowing
Background EEG slowing	Present	Diffuse	Common aging biomarker
EEG reactivity	Decreased	General	Weaker responses to sensory stimuli
Integrative brain activity	Decreased	Global cortical networks	Lower cross-regional coordination
EEG wave regularity	More irregular	Diffuse	EEG waveforms are less stereotyped and more variable
Global EEG network	Inefficient	Global	With aging, the global EEG network becomes increasingly fragmented, random, and inefficient

EEG: electroencephalography.

EEG connectivity and network changes with aging

Aging alters the functional and structural network organization of the brain in several ways. Across studies using EEG-based graph-theoretical analyses, aging is consistently associated with reduced global network efficiency, weakened local connectivity, greater network randomness, and reorganization of the hub structure (Fig. 4B) [103–109]. Multiple studies support this argument, reporting a negative correlation between global efficiency (a measure of how effectively information is integrated across distant brain regions) and age as well as a positive correlation between characteristic path length (the average number of steps required for information to travel between nodes) and age [103–105]. These findings indicate that information transfer slows and becomes less coordinated in older adults, reflecting weakened large-scale brain integration. At the local level, clustering coefficients decrease with age, suggesting that nearby regions (modules) become less tightly interconnected and communicate less efficiently than in younger brains [103,106–109]. Thus, the aging process disrupts the brain’s small-world architecture, which is a network organization that balances local specialization and global integration for optimal information processing. Decreased clustering and increased path length collectively indicate a shift toward a more random network topology, reflecting the loss of both segregation and integration [108,110].

Another robust finding is a decline in modularity, which is the degree to which the brain network is segregated into distinct communities [103,107,109]. Older adults show reduced differentiation of network modules, particularly in higher-frequency bands (beta and gamma) [103,107,109]. This decline in modular organization reflects a loss of functional specialization, which, in turn, reduces cognitive flexibility. In addition, aging alters the hub structure of the brain network. Posterior hubs, such as those in the occipital and parietal regions, tend to weaken, whereas frontal hubs often become more prominent [107,111]. This shift toward frontal connectivity likely reflects a compensatory mechanism in which the aging brain reroutes communication to maintain function despite reduced network efficiency.

Changes in brain connectivity with age can also be visualized using an analogy to the global airline system, where airports represent brain regions (nodes) and flight routes correspond to functional connections (edges). In a young, healthy brain, this system operates as a small-world network, similar to an optimally designed air traffic map that balances global and local routers. Major hub airports such as New York, London, or Tokyo connect efficiently across long distances, enabling rapid global communication, whereas smaller regional airports handle short-range routes for specialized, localized processing. This organization maintains an ideal balance between local specialization and global integration, ensuring resilience and efficiency. By contrast, the aging brain resembles a disrupted and increasingly random air traffic network. Long-distance routes weaken or disappear (reflecting reduced long-range connectivity and global efficiency); regional airports lose coordination (lower local clustering); and flights are rerouted through longer, inefficient paths (increased path length). Major hub airports lose influence as traffic becomes unevenly distributed (hub reorganization).

Individual hypnotic agents and associated EEG features in the elderly

Propofol

At the point of propofol-induced loss of consciousness, elderly patients have recently been found to exhibit reduced power across the entire EEG frequency spectrum compared to younger adults [112], a finding corroborated by an earlier study [113]. Notably, during the maintenance of anesthesia with propofol, a reduction in alpha power (8–12 Hz) and coherence has been observed in elderly patients (Fig. 8), which aligns with age-related changes in thalamocortical function [113]. Furthermore, elderly patients are more likely to exhibit burst suppression, even after a single bolus dose of propofol equivalent to that administered to younger individuals [21,113], suggesting increased sensitivity to propofol with age, which may lead to deeper anesthetic states at lower doses. Near the point of loss of consciousness in elderly patients, both the bispectral index (BIS) and Lempel-Ziv complexity, an EEG-derived measure of signal complexity, have been reported to be higher than those in younger adults, suggesting the presence of age-related bias in these indices [114]. Indeed, age-related bias in BIS measurements during propofol- and sevoflurane-based anesthesia has been previously reported [115,116]. Given that permutation entropy has been reported to exhibit relatively minimal age-related bias [114], the development of age-independent commercial indices may represent a clinically meaningful advancement.

Fig. 8.

Age-related differences in electroencephalography (EEG) spectrograms during propofol infusion in healthy volunteers. (A) EEG spectrogram of a 45-year-old volunteer during continuous propofol infusion. (B) spectrogram of a 70-year-old volunteer under comparable infusion conditions. In the younger subject (A), prominent alpha (8–12 Hz) and delta (1–4 Hz) oscillations are evident. In contrast, in the older subject (B), alpha oscillations are markedly attenuated, while delta activity remains present. These findings illustrate the age-related attenuation of alpha activity during propofol infusion.

Inhaled anesthetics

Similar to propofol, a decrease in EEG power across all frequency bands with advancing age has been observed during sevoflurane anesthesia [113]. While slow oscillations (0.1–1 Hz) are preserved, alpha oscillations (8–12 Hz) decline markedly with age [113]. Notably, the alpha-to-slow ratio is significantly lower in older adults than in younger individuals, primarily due to a disproportionately greater reduction in alpha-band power relative to slow-band power [113]. In addition, alpha oscillations tend to become progressively slower with age [117]. Theta oscillations, which are typically prominent in younger adults at sevoflurane concentrations exceeding 1 MAC, are often absent or markedly diminished in elderly patients [118]. In addition, EEG features in the elderly tend to exhibit a more irregular oscillatory composition than those in younger individuals [119]. Moreover, the likelihood of burst suppression at equivalent end-tidal sevoflurane concentrations is significantly higher in older individuals than in their younger counterparts [113].

Dexmedetomidine

Because dexmedetomidine is primarily used as an adjunct to general anesthesia or for sedation, studies directly comparing EEG characteristics between elderly and younger individuals are limited, making it difficult to clearly describe age-specific EEG features of dexmedetomidine. In a study involving patients aged > 70 years undergoing anesthesia with propofol and remifentanil, those who received additional dexmedetomidine exhibited significantly lower power in the alpha and beta bands than those who received propofol alone [120]. However, this finding should be interpreted with caution, as dexmedetomidine was co-administered with other agents and the propofol concentration was intentionally reduced in the dexmedetomidine group to maintain an appropriate depth of anesthesia. Therefore, the observed EEG changes cannot be solely attributed to the effects of dexmedetomidine. In another study using mice aged 4–5 months (young adult) and 10–18 months (aged), dexmedetomidine was administered in a dose-dependent manner to induce sedation [121]. High-voltage spike waves resembling epileptiform activity were observed at higher doses and occurred significantly more frequently in aged mice [121].

Ketamine

Relatively few studies have examined the EEG characteristics of ketamine in the elderly. One study investigated intraoperative EEG changes following adjunctive ketamine administration during anesthesia maintenance in patients aged ≥ 65 years who underwent spinal surgery [122]. In patients with normal preoperative cognitive function, ketamine administration led to a significant increase in power within the 10–20 Hz range. However, this change was not observed in patients with pre-existing cognitive impairment [122], suggesting that EEG responses to ketamine may differ depending on the cognitive status of older adults. One limitation of this study is that the EEG data were recorded from the frontal cortex using a SedLine Root monitor (Masimo Corporation), which has a relatively low sampling rate. As a result, only frequencies between 0.5 and 30 Hz were analyzed; thus, any occurrence of ketamine-induced increases in gamma band (≥ 30 Hz) power could not be analyzed. In an experimental study using rats aged 2–3 months (young adult) and 15–17 months (aged), intraperitoneal administration of ketamine resulted in a significant increase in gamma band (20–100 Hz) power in both age groups [123]. However, aged rats exhibited an increase in relative power in the delta band (2–4 Hz) and a decrease in the theta band (4–8 Hz), indicating age-related differences in EEG spectral composition [123].

Remimazolam

In a study comparing EEG characteristics during anesthesia induction between propofol and remimazolam in patients aged ≥ 65 years, remimazolam exhibited EEG patterns similar to those of propofol [90]. Following loss of consciousness, both alpha and delta oscillations were observed, and a greater reduction in feedback connectivity than in feedforward connectivity resulted in disrupted asymmetry in frontoparietal connectivity, consistent with findings reported for propofol [90]. However, the absolute value of the normalized symbolic transfer entropy, a metric reflecting the amount of directional information flow, was lower than that observed in younger adults in a previous study [53]. This may be interpreted as resulting from a general reduction in the overall information content of the elderly brain. In another study comparing the EEG features of remimazolam and midazolam during anesthesia induction in patients aged > 65 years, remimazolam was associated with significantly higher alpha band activity in the frontal cortex than midazolam [124].

Conclusion

Aging is an inevitable physiological process that affects the brain, resulting in characteristic changes in EEG activity. In elderly individuals, a general decline in EEG power is observed across all frequency bands, including the alpha band. This is accompanied by the slowing of background rhythms and a reduction in EEG reactivity to external stimuli. These global alterations reflect age-related deterioration in neuronal function and connectivity. Within this overarching framework of EEG slowing and power reduction, the brain’s response to various hypnotic agents undergoes notable changes. Compared with younger adults, elderly patients often exhibit distinct EEG patterns during sedation and anesthesia, such as reduced alpha power, earlier onset of burst suppression, or altered oscillatory dynamics, depending on the specific agent used. The network structure of the brain becomes less globally integrated and increasingly fragmented into local subnetworks. Some brain regions form connections in a more random and inefficient manner, reducing overall network efficiency. As a result, communication between distant regions becomes slower, less reliable, and more energy-consuming. Although the brain remains functional, its network operates with reduced efficiency, flexibility, and resilience, mirroring the cognitive slowing and diminished adaptability observed during aging. However, the degree and nature of these changes can vary significantly between individuals and are influenced by factors such as cognitive reserve, comorbidities, and pharmacodynamic sensitivity. A comprehensive understanding of age-related EEG alterations, both in baseline physiology and in response to anesthetic agents, is essential for optimizing care in older adults. By recognizing and interpreting these changes appropriately, clinicians can better tailor anesthetic strategies to minimize risks and ensure safe and effective management of elderly patients undergoing surgery or procedural sedation.