Age-related alterations in the EEG spectral composition can bias the outputs of pEEG algorithms. The earliest study, by Kratzer et al. [10], examined propofol-induced BS and showed that aging is characterized by diminished alpha power and elevated permutation entropy (PE), which is indicative of greater temporal complexity. This result implies that complexity-based pEEG measures such as PE can paradoxically increase with age despite deep anesthesia, potentially complicating the detection of BS in older patients. Obert et al. [11] analyzed EEG data from 180 adults (aged 18–90 years) under sevoflurane anesthesia using multiple commercial algorithms. BIS, quantitative consciousness index (qCON), state entropy (SE), and response entropy (RE) increased linearly with age (approximately 2 index units/decade), spectral edge frequency (SEF) rose by approximately 0.5 Hz/decade, Treaton decreased, and Narcotrend showed no age effect. Obert et al. [12] further reported that PSI increased by approximately 2 index points/decade regardless of the anesthetic agent, with SEF showing a similar upward trend. Biggs et al. [13] compared EEG parameters in younger and older adults during propofol induction and noted a higher BIS, Lempel–Ziv complexity, and SEF in older adults, accompanied by reduced alpha power. More recently, Ebensperger et al. [14] conducted a large-scale analysis of over 14,000 adult cases and confirmed that SE and RE increase with age during steady-state anesthesia, despite equivalent or higher age-adjusted minimum alveolar concentration (MAC) values. Even within the “adequate” SE range (40–60), older patients frequently exhibited BS, and the likelihood of this mismatch rose sharply with age (ρ = 0.90), highlighting the risk of inadvertent anesthetic overdose if age-related EEG changes are not considered.

Collectively, these studies suggest that age-related changes in EEG, characterized by marked attenuation of alpha and slow–delta activity with relative preservation of higher frequencies and increased temporal irregularity, are expressed differently across pEEG metrics depending on their underlying algorithms. The spectral entropy indices (SE and RE) primarily increase owing to a flatter power spectrum and even spectral distribution [11,14]. Complexity metrics (PE and Lempel–Ziv complexity) rise with greater temporal pattern variability and reduced rhythmicity [10,13]. Composite indices (BIS, PSI, and qCON) also tend to increase with age, most likely reflecting spectral redistribution [11-13].

In pediatric patients, the reliability of pEEG indices varies markedly with age, reflecting the rapid and nonlinear maturation of EEG features in early childhood (Table 1, Fig. 1A). The late emergence and gradual maturation of frontal alpha oscillations, a crucial component of many pEEG algorithms, indicate that before full development, indices such as the BIS and PSI do not accurately reflect anesthetic depth, particularly in infants and toddlers. Unlike the BIS, which is based solely on adult EEG data, the PSI incorporates pediatric-specific algorithms; however, its reliability in infants and young children remains uncertain.

Tokuwaka et al. [15] examined MAC_BIS50 (end-tidal sevoflurane at which BIS would be 50), but were unable to determine it in 1-year-old children, as BIS never decreased below 50, even at concentrations up to 4.8% sevoflurane. In a prospective trial of 111 children aged 1–18 years, Ricci et al. [16] found that children aged < 2 years consistently exhibited higher PSI values than older children, despite similar age-adjusted dosing. Sciusco et al. [17] analyzed 48 children across three age groups (infants, toddlers, and school-aged children) and demonstrated that the concentration–index coupling strengthened as they grew, with BIS and SE values being lowest in infants and highest in toddlers.

In children aged 2–6 years, alpha power increased; however, spectral variability remained high. In a cohort of children aged 6 months to 12 years, Kim et al. [18] reported that the BIS and PSI correlated only fairly (r = 0.43), with both indices being weakly associated with age-adjusted end-tidal sevoflurane concentrations, particularly in children aged 2–7 years. Jang et al. [19] further demonstrated that while the BIS and PSI could discriminate state transitions (maintenance vs. emergence), absolute values correlated poorly with sevoflurane concentration.

Overall, in children, especially those aged < 6–7 years, absolute pEEG thresholds should not be used in isolation. Because age-specific calibration is lacking and interindividual variability is large, interpretation is intrinsically challenging. Hence, trend tracking, along with raw EEG and spectrogram reviews, is essential for accurate depth assessment.

The concept of pEEG-guided anesthesia as a strategy to reduce POD in older surgical patients has been extensively tested. The ENGAGES trial enrolled 1,232 adults aged ≥ 60 years undergoing major noncardiac surgery to test whether BIS-guided anesthesia with targeting BIS ≥ 40 and minimizing BS would reduce POD [20]. Although the intervention group received lower end-tidal anesthetic concentrations and had a shorter BS duration, the incidence of POD was not significantly reduced. In contrast, a BALANCED trial substudy involving 515 patients aged ≥ 60 years compared light anesthesia (target BIS 50) with deep anesthesia (target BIS 35). They found that the light anesthesia group had a significantly lower incidence of POD and less cognitive decline 1 year postoperatively, suggesting potential long-term benefits of avoiding excessive anesthetic depth [21]. A meta-analysis of nine RCTs, including these large multicenter trials, concluded that pEEG-guided anesthesia or intentionally lighter depth management modestly reduces POD risk compared with standard care [22]. The incidence of POD in the pEEG-guided or lighter-target group was 19.0% (440/2,310) compared with 23.3% (545/2,338) in usual care or deeper targets (pooled odds ratio = 0.78; 95% confidence interval, 0.60–1.00; P = 0.054). Although the effect size was small, accumulating evidence suggests that pEEG index-guided anesthesia contributes to reducing POD by avoiding unnecessary deep anesthesia and prolonged BS in older patients.

In children, randomized evaluations of pEEG-guided anesthesia have produced inconsistent effects on the emergence phenomena, even when dosing becomes more disciplined under guidance. The earliest trial, reported in 2010, randomized children aged 2–12 years undergoing propofol total intravenous anesthesia (TIVA) to either BIS-guided care (targeting 45–60) or routine practice. They found no reduction in EA or improvement in the recovery profile without a difference in propofol amount [23]. In a subsequent trial of children aged 2–8 years undergoing ophthalmic surgery under sevoflurane anesthesia, BIS targets of 40–45 (deep) and 55–60 (light) were compared [24]. The study found that lighter targeting did not significantly reduce EA, underscoring that merely driving a higher index does not translate into better emergent behavior. In a randomized trial of 375 children aged 1–3 years undergoing propofol–remifentanil TIVA, BIS-guided care (target 45–60) did not shorten extubation time or post-anesthesia care unit (PACU) stay compared with standard practice, and propofol consumption was higher [25]. In contrast, a randomized, double-blind trial in 163 preschool children aged 3–8 years undergoing endoscopic adenoidectomy with sevoflurane found that BIS-guided anesthesia (target 40–60) reduced the incidence of emergence delirium (ED) and produced lower pediatric anesthesia emergence delirium (PAED) scores with lower volatile exposure than standard care with maintenance of 1–1.2 MAC [26]. The most recent randomized trial in children aged 4–18 years comparing BIS-guided anesthesia (target 45–60) with standard care without pEEG monitoring showed a lower incidence of EA, greater readiness for discharge, and consistently lower volatile anesthetic exposure [27].

Overall, the effects of BIS-guided anesthesia in children are inconsistent. A reduction in EA was reported in a single-center trial, whereas several RCTs, including a recent multicenter study, found no difference between EA and standard care. Similarly, anesthetic consumption is mixed; some studies have observed lower end-tidal sevoflurane concentrations with BIS guidance, while others have reported no reduction or even higher propofol use in TIVA cohorts. Overall, routine BIS guidance in pediatric anesthesia cannot be recommended at present, although future trials may better define its role in improving recovery (Table 3).

Multiple intraoperative EEG parameters have been reported as potential predictors of POD; however, most have not been robustly validated. Thus far, BS and alpha oscillations have been established as complementary intraoperative EEG markers of brain vulnerability.

BS represents profound cortical inactivation that occurs more readily and for longer durations in older or cognitively vulnerable patients, plausibly disrupting thalamocortical communication during the perioperative period [3-5]. It has been extensively studied in relation to POD. In a recent meta-analysis of 14 studies (n = 6,435), intraoperative BS was associated with higher odds of POD (odds ratio 1.49) [28]. Continuous measures (e.g., suppression ratio or cumulative suppression time) were more informative than binary present/absent definitions. These data support the minimization of unnecessary deep anesthesia and exposure to BS, particularly in vulnerable older adults, and favor monitoring strategies that track the suppression burden rather than using index thresholds alone.

Frontal alpha power during GABAergic anesthesia reflects the capacity of thalamocortical circuits to generate coherent rhythmic activity [29]. The attenuation of this signal has repeatedly been associated with adverse cognitive trajectories in clinical research. In an observational cohort of older surgical patients, those who developed postoperative subsyndromal delirium showed markedly lower absolute and relative alpha powers during maintenance anesthesia, and the relative alpha fraction discriminated the risk with high accuracy [30]. These associations persisted after accounting for the anesthetic dose (i.e., not explained by deeper anesthesia). Similarly, Kinoshita et al. [31]’s observational study of 80 patients undergoing highly invasive cancer surgery under propofol TIVA demonstrated that a lower perioperative relative alpha ratio measured up to 21 h after anesthesia induction was independently associated with POD, with a greater predictive value than inflammatory markers. In a more recent prospective study of 151 patients, Pollak et al. [32] showed that reduced frontal alpha power could identify patients at risk of POD within 1 min of loss of consciousness. Finally, in the pooled analyses of three prospective cohorts (n = 82), the dose-adjusted intraoperative alpha power independently predicted POD and slower postoperative processing speed [33]. Overall, diminished alpha power appears to be a reproducible marker of POD risk in older adults, and multiple studies have indicated that this association is not solely explained by anesthetic depth.

Beyond simple prediction, several mediation analyses have indicated that intraoperative EEG phenotypes transmit part of the effect of preexisting vulnerability to POD. In a retrospective substudy of the MINDDS trial of older patients (aged > 60 years) undergoing cardiac surgery (n = 159), BS occurring during cardiopulmonary bypass (CPB) mediated the relationships between lower preoperative physical function, lower nadir CPB temperature, reduced intraoperative alpha power, and POD. This finding supports BS as a mechanistic pathway rather than a mere correlate [34]. In the secondary analysis of the ENGAGES trial (n = 1,113), preoperative cognitive abnormalities increased POD risk, and a small but statistically significant portion of this effect was transmitted through intraoperative EEG suppression [35]. Randomization to EEG-guided care did not materially change the mediating effect. In a prospective study of 252 older patients aged 65 years or older receiving noncardiac surgery under propofol TIVA, cumulative EEG suppression time mediated approximately 24% of the association between preoperative frailty and delirium severity. This result further implicates suppression as a causal conduit from physiological vulnerability to POD [36]. More recently, a prospective study of 138 orthopedic patients aged 60 years or older receiving propofol TIVA demonstrated that diabetes was associated with up to 3.2 times higher POD odds and low intraoperative frontal alpha power partially mediated this relationship (approximately 20%). This finding suggests that deficient alpha generation under anesthesia contributes to delirium risk in metabolically vulnerable brains [37]. Collectively, these analyses support a mechanistic model in which excessive cortical inactivation and attenuated anesthetic-induced alpha oscillations carry part, but not all, of the risk conferred by physical and cognitive impairments that increase susceptibility to POD. These findings underscore the importance of minimizing unnecessary suppression and preserving intraoperative alpha activity while also addressing upstream vulnerabilities such as frailty and cognitive impairment.

Evidence of pediatric EEG predictors of EA remains limited and heterogeneous. In a study of 97 children aged 0.5–8 years undergoing strabismus surgery with propofol or sevoflurane induction and sevoflurane maintenance, Koch et al. [38] reported that neither the occurrence nor the duration of BS was associated with EA. A further trial examining the association of multiple BS metrics under anesthesia with postoperative awakening delirium and other adverse outcomes in a similar age group is ongoing [39].

Spectral analyses of emergence yielded mixed results. In a study of 2–10-year-old children recovering from sevoflurane anesthesia, Kim et al. [40] reported that higher frontal delta and lower alpha–beta power correlated with greater EA severity. In contrast, a 2016 randomized trial of 29 children aged 1–6 years found that greater alpha power and lower theta power during emergence from sevoflurane anesthesia were associated with a higher incidence of EA [41].

Overall, in children under 10 years of age, neither intraoperative EEG suppression nor alpha-band changes at emergence could be used as standalone predictors of EA.

Given the emerging links between intraoperative EEG suppression/alpha attenuation and postoperative neurocognitive disorders, real-time interpretation of spectrograms and raw EEG has been proposed as an alternative to relying solely on pEEG indices. This method would enable anesthesiologists to promote desirable patterns, such as sustained frontal alpha activity, while avoiding maladaptive features such as BS or excessive slowing.

In a recent retrospective study of 140 adults undergoing elective craniotomy with propofol-based TIVA, spectrogram-guided titration, aimed at sustaining robust frontal alpha power while avoiding BS, was associated with lower anesthetic consumption, fewer delayed-emergence events, and better in-hospital functional status compared with BIS-guided care (target 40–60). However, delirium outcomes were unchanged [42]. Gaskell et al. [43] proposed a protocol for a randomized trial that would test an explicit alpha-maintenance strategy through targeted volatile anesthetic titration with adjunct opioid boluses against SE or BIS-guided standard care, with PACU delirium and early cognitive recovery as the planned outcomes. He et al. [44] conducted a randomized trial in 125 adults aged 60 years or older using raw EEG guidance to maintain a stable slow–delta pattern rather than targeting alpha activity. They found no reduction in POD compared with the standard titration based on vital signs. Even in that trial, reduced frontal alpha power was independently associated with delirium, emphasizing its potential mechanistic relevance despite not being a treatment target.

Pediatric data are also mixed. A randomized trial in boys aged 2–12 years has shown that sevoflurane anesthesia targeting 10–15 Hz SEF with strict avoidance of BS reduced the occurrence of ED and improved the recovery profile with lower volatile exposure than standard clinical titration [45]. In contrast, a randomized trial of 190 children aged 1–6 years who titrated sevoflurane to maintain continuous slow–delta oscillations and avoid suppression (with PSI 25–50 when feasible) achieved lower end-tidal concentrations and reduced EEG suppression. However, no difference was observed in PAED score or other recovery profiles [46]. A pediatric protocol is currently underway to evaluate whether maintaining an alpha-/delta-dominant frontal pattern and avoiding suppression reduces EA and improves recovery metrics [47].

Overall, the spectrogram- and raw EEG-guided approaches are feasible and biologically motivated. Adult and pediatric trials suggest anesthetic-sparing effects, whereas the effects on postoperative neurocognitive disorders remain heterogeneous and likely depend on the specific EEG, patient age, and surgical context. Table 4 summarizes available adult and pediatric studies.