Quantitative Electroencephalography Analysis in Panic Disorder: Exploring the Neurophysiological Significance of High Beta Activity

Abstract

Objective

This study aimed to investigate the neurophysiological characteristics of panic disorder (PD) by analyzing quantitative electroencephalography (QEEG) data, with a particular focus on high beta activity.

Methods

In this retrospective study, resting-state QEEG data from 58 patients with PD and 23 healthy controls (HC) were analyzed. Spectral power was calculated for delta (1–4 Hz), theta (4–8 Hz), alpha (8–12 Hz), beta (12–25 Hz), and high beta (25–30 Hz) bands across frontal, central, and posterior regions. Group differences were assessed using generalized estimating equations, and Pearson correlation analyses were conducted to examine associations between electroencephalography activity and clinical symptoms.

Results

PD patients showed significantly higher high beta power across all regions compared to HC (frontal: estimate=1.276, χ²=12.48, p<0.0010; central: estimate=0.874, χ²=9.87, p=0.0017; posterior: estimate=0.524, χ²=4.48, p=0.0343). Beta power was elevated only in the frontal region (estimate=3.391, χ²=5.31, p=0.0212), while delta power was increased overall (χ²=4.60, p=0.0390) without regional specificity. No significant group differences were observed in alpha or theta bands. A significant positive correlation was found between frontal high beta power and Panic Disorder Severity Scale scores (r=0.41, p=0.010), while no other significant correlations were observed between regional high beta power and clinical scales.

Conclusion

Increased high beta, beta, and delta activity in PD may reflect a neurophysiological imbalance. In particular, elevated high beta power may indicate hyperarousal and excessive cognitive control in PD, suggesting its potential as a future biomarker.

INTRODUCTION

Panic disorder (PD) is a chronic anxiety disorder characterized by recurrent, unexpected episodes of intense fear or anxiety. These panic attacks are accompanied by various autonomic nervous system symptoms, including palpitations, sweating, trembling, shortness of breath, chest discomfort, nausea, dizziness, and fear of death [1]. Due to persistent anticipatory anxiety about future attacks, patients commonly avoid specific places or situations. As a result, these symptoms significantly impair patients’ daily lives, social and occupational functioning, leading to a marked reduction in quality of life [2]. Epidemiological studies estimate the lifetime prevalence of panic attacks of 28.3%, with 4.7% meeting the diagnostic criteria for PD. The disorder primarily affects young adults in their 20s and 30s, a period of peak social and occupational activity, with women showing 2–3 times higher prevalence than men [3]. Additionally, PD frequently co-occurs with other psychiatric conditions such as major depression, generalized anxiety disorder, and social phobia [4].

Research on the neurobiological mechanisms of PD has focused on abnormalities within the fear network. In particular, dysfunctional sensory information processing between the cortex and brainstem is recognized as a key pathophysiological feature [5]. This network involves interconnected structures, including the prefrontal cortex (PFC), hippocampus, and amygdala, which play a crucial role in processing emotions such as fear and anxiety. Hyperactivation of the amygdala stimulates the hypothalamic-pituitary-adrenal axis, increasing cortisol secretion, and amplifies autonomic nervous system responses through the periaqueductal gray (PAG), locus coeruleus, and parabrachial nucleus, leading to core symptoms of panic attacks such as tachycardia and hyperventilation [6-8]. While PFC typically inhibits excessive amygdala activity to mitigate anxiety and fear, this regulatory function appears to be impaired in PD patients [6,9,10].

Although our understanding of the pathophysiological mechanisms underlying PD has evolved, clinical diagnosis still heavily relies on patients’ subjective symptom reports and psychological assessments [11]. This reliance on subjective measures limits the objectivity and reliability of diagnoses. In this context, electroencephalography (EEG) is gaining attention as a non-invasive tool capable of measuring real-time brain electrical activity. In particular, quantitative electroencephalography (QEEG) objectively evaluates changes in brain activity across various frequency bands by quantifying EEG data through mathematical algorithms like Fourier transform [12]. This approach offers objective and quantifiable biomarkers to support psychiatric diagnosis, complementing current subjective assessments [13,14].

Currently, the clinical utility of QEEG has been well established in attention-deficit/hyperactivity disorder [15], and it is increasingly being applied to other psychiatric conditions. Recent reports suggest that QEEG may also show characteristic findings in post-traumatic stress disorder, including increased gamma and delta activity [16]. However, research in PD remains relatively limited, making it difficult to identify consistent findings [17,18]. Studies on QEEG in PD face methodological challenges, as the unpredictable nature of panic attacks makes it challenging to directly measure EEG changes during an attack. Given that patients with PD often experience persistent anticipatory anxiety and chronic arousal even between attacks, understanding changes in brain activity in the resting state holds significant importance.

Although studies on QEEG in PD remain limited, existing findings have generally reported increased beta activity, decreased alpha activity, and increased delta activity in the frontal regions. Increased beta activity and decreased alpha activity were thought to reflect possible deficiency in top-down regulatory control over anxiety. Additionally, increased beta activity reflected a state of high excitability, anticipatory anxiety and maintenance of a hypervigilant cognitive state. A paradoxical shift towards slow wave delta activity was thought to reflect altered brainstem evoked response and physiological hyperarousal [19-25].

Among these oscillatory changes, beta frequency wave (12–30 Hz) warrants particular attention because it encompasses sub-bands that differ in their functional roles. Beta waves (12–25 Hz) are known to play a crucial role in maintaining attention, behavior control, and cognitive regulation [26], whereas high beta activity (25–30 Hz) has been associated with heightened stress, hyperarousal, and anxiety [27,28]. In healthy individuals, exposure to fear-inducing stimuli has been shown to elicit increased high-beta activity in the frontal region [28]. Clinically, elevated high beta activity has been observed in various anxiety-related conditions, including generalized anxiety disorder and phobias, as well as in major depressive disorder comorbid with prominent anxiety symptoms [29,30]. This functional distinction between beta sub-bands may be relevant for understanding PD.

However, most QEEG studies on PD have analyzed beta waves (12–30 Hz) as a single frequency band [17,20,25], which may have obscured the potential differences between arousal-related high-beta activity and regulatory beta activity. Given that PD is characterized by persistent autonomic hyperarousal and impaired prefrontal regulation, distinguishing between these beta frequency bands may provide a more nuanced understanding of the neurophysiological mechanisms of PD.

Therefore, this study aimed to explore the QEEG characteristics of patients with PD in a resting state, with a particular focus on differentiating beta and high beta waves to better elucidate the role of high beta activity in PD. Based on previous studies reporting elevated beta wave activity in PD [17,20,25], this study hypothesized that both beta and high beta activity would be elevated in the PD group compared to the healthy controls (HC) group, with a more pronounced increase in high beta activity. Additionally, changes in alpha, theta, and delta waves were analyzed to provide a more comprehensive understanding of neural activity associated with PD.

METHODS

Participants

Patients between 18 and 65 years of age who visited the Department of Psychiatry at Seoul Metropolitan Government‒ Seoul National University (SMG‒SNU) Boramae Medical Center from January 2020 to December 2023 with a diagnosis of PD were selected. Patients who underwent QEEG were included for retrospective analysis, and their psychological test results were analyzed when available. To investigate the neurophysiological characteristics of pure PD, we specifically excluded patients with common psychiatric comorbidities such as major depressive disorder, and those with seizure disorders, history of brain injury, intellectual disabilities, or psychotic disorders.

Additional recruitment of HC and PD patients was conducted through online advertisements (https://www.alllivec.co.kr/) SMG‒SNU Boramae Medical Center bulletin board. All participants underwent interviews by psychiatrist following Diagnostic and Statistical Manual of Mental Disorders-Fifth Edition diagnostic criteria, and standardized clinical scales including the Panic Disorder Severity Scale (PDSS), Beck Anxiety Inventory (BAI), and Beck Depression Inventory (BDI) were administered to assess symptom severity [31-33]. The final study population comprised 58 PD patients and 23 HC.

This study was conducted in accordance with the Declaration of Helsinki (revised 2013) and ICH-GCP guidelines. The study protocol and participant rights protection were reviewed and approved by the Institutional Review Board SMG‒SNU Boramae Medical Center (IRB 30-2024-48).

EEG recording and preprocessing

Participants underwent EEG recording while lying in a comfortable chair in a dimly lit, electrically shielded room. EEG was recorded for 5 minutes with eyes closed using a 19-channel electrode cap positioned according to the international 10–20 system. Linked ear reference montage was used, with the ground electrode placed between FPz and Fz electrodes. Analog signals from the scalp were digitized at 500 Hz through an Analog-to-Digital converter, with high-pass and low-pass filters set at 1 Hz and 40 Hz, respectively.

Recording equipment consisted of NeuroScan SynAmps2 (Scan 4.5; Compumedics), and data analysis was performed using NeuroGuide software (version 2.6.1; Applied Neuroscience). Initial 300 epochs were recorded from which 60 artifact-free epochs were selected for analysis following thorough visual inspection using NeuroGuide’s artifact rejection tools.

The preprocessed EEG data were Fourier transformed to obtain absolute power (μV²), relative power (%), and peak frequency (Hz) for delta (1–4 Hz), theta (4–8 Hz), alpha (8–12 Hz), beta (12–25 Hz), and high beta (25–30 Hz) frequency bands. The 19 electrodes were grouped into three regions by averaging: frontal (FP1, F3, F7, Fz, FP2, F4, F8), central (T3, C3, Cz, T4, C4), and posterior (T5, P3, O1, Pz, T6, P4, O2) [34].

Statistical analysis

Prior to conducting statistical analyses, data were reviewed to identify and remove outliers that could potentially skew results. To account for correlations among multiple measurements from various brain regions, generalized estimating equations (GEE) were employed. GEE is an extension of generalized linear models capable of handling multivariate responses and has been validated in numerous EEG data analyses for evaluating group effects [34,35].

In this study, GEE was used to assess the effects of group (PD and HC group) and brain region (frontal, central, posterior) and their interactions. When group-by-region interactions were significant, absolute power differences between groups were analyzed for each brain region. When interaction was not significant, only group effect was analyzed. All GEE analyses were adjusted for age and sex.

Demographic data and clinical scales between PD and HC groups were compared using independent sample t-tests or chi-square tests. Additionally, Pearson’s correlation coefficients were used to explore relationships between QEEG absolute power and clinical scales. Although clinical scales (BDI, BAI, and PDSS) showed significant differences between groups, these measures were excluded from the multivariable GEE models to avoid multicollinearity, since these clinical scores are strongly correlated with group status.

All statistical analyses were performed using IBM SPSS Statistics version 29 (IBM Inc.) and R version 4.4.2. P-values below 0.05 were considered statistically significant. However, correlation analyses were conducted for exploratory purposes and thus p-value correction for multiple comparisons were not applied.

RESULTS

Demographic and clinical data

Statistical analyses were conducted on data from PD (n=58) and HC (n=23). As shown in Table 1, the mean age of the PD group was 38.41 years (SD=12.96), significantly higher than the HC group’s mean age of 28.65 years (SD=7.80) (t=-4.148, p<0.001). Chi-square testing showed no significant difference in diagnosis rates between males and females (χ²=0.754, p=0.384).

Table 1.

Demographic and clinical characteristics of study participants (N=81)

	HC (N=23)	PD (N=58)	χ² or t	p
Sex (male/female)	6/17	21/37	0.754	0.384
Age (yr)	28.65±7.80	38.41±12.96	-4.148	<0.001***
BDI	6.61±4.71	22.61±13.36	-6.940	<0.001***
BAI	3.09±3.23	23.95±13.53	-9.083	<0.001***
PDSS	0.00±0.00	14.74±6.62	-13.916	<0.001***

Data are presented as mean±standard deviation or number (%).

^***p<0.001.

PD, panic disorder; HC, healthy controls; BDI, Beck Depression Inventory; BAI, Beck Anxiety Inventory; PDSS, Panic Disorder Severity Scale.

All HC group participants completed clinical scales. In the PD group, 41 participants completed the BDI, 38 completed the BAI, and 39 completed the PDSS. The PD group showed significantly higher scores on all clinical scales: BDI (22.61±13.36 vs. 6.61±4.71), BAI (23.95±13.53 vs. 3.09±3.23), PDSS (14.74±6.62 vs. 0.0±0.0), all comparisons p<0.001.

EEG activity (absolute power)

Figure 1 shows topographic maps comparing group-averaged absolute amplitude values for delta, theta, alpha, beta, and high beta bands between PD and HC groups, providing a visual representation of EEG activity distribution across the scalp. The PD group showed increased high beta power across all regions, increased frontal beta power, and higher overall delta power (not region-specific) compared to HC.

Figure 1.

Topographical maps of absolute power in the PD patient and HC groups. Scales show μV² for absolute power. Red represents higher values, and blue represents lower values. The PD group showed increased high beta power across all regions, increased frontal beta power, and higher overall delta power (not region-specific) compared to HC. PD, panic disorder; HC, healthy controls.

Table 2 summarizes the GEE analysis results regarding distinct patterns in specific frequency bands between PD and HC group. For high beta band, analysis showed a significant group-by-region interaction effect (χ²=17.19, p<0.0010). In subsequent regional comparisons presented in Table 3, the PD group demonstrated higher absolute power across all brain regions, with the most pronounced difference in the frontal region (estimate=1.276, χ²=12.48, p<0.0010), followed by the central (estimate=0.874, χ²=9.87, p=0.0017) and posterior regions (estimate=0.524, χ²=4.48, p=0.0343).

Table 2.

GEE analysis of group and brain region effects on EEG absolute power

Absolute power (μV²)	χ²	df	p
Delta
Group	4.64	1	0.0312*
Region	73.32	2	<0.0010***
Group×Region	0.80	2	0.6708
Group (no interaction)	4.60	1	0.0390*
Theta
Group	0.34	1	0.5590
Region	63.50	2	<0.0010***
Group×Region	0.32	2	0.8520
Group (no interaction)	0.35	1	0.5561
Alpha
Group	2.13	1	0.1441
Region	12.71	2	0.0017**
Group×Region	6.06	2	0.0483*
Beta
Group	0.93	1	0.3357
Region	7.34	2	0.0254*
Group×Region	6.28	2	0.0433*
High beta
Group	4.48	1	0.0343*
Region	13.13	2	0.0014**
Group×Region	17.19	2	<0.0010***

In the absence of Group×Region interaction effect, the group effects alone were also analyzed shown as group (no interaction).

^*p<0.05;

^**p<0.01;

^***p<0.001.

GEE, generalized estimating equations; EEG, electroencephalography.

Table 3.

Regional comparisons of alpha, beta, and high beta EEG power following significant group-by-region interaction

Absolute power (μV²)	Estimate	χ²	p
Alpha
Frontal	0.644	0.02	0.9023
Central	-2.202	0.21	0.6483
Posterior	-11.268	2.13	0.1441
Beta
Frontal	3.391	5.31	0.0212*
Central	1.550	0.86	0.3545
Posterior	1.738	0.93	0.3357
High beta
Frontal	1.276	12.48	<0.0010***
Central	0.874	9.87	0.0017**
Posterior	0.524	4.48	0.0343*

^*p<0.05;

^**p<0.01;

^***p<0.001.

EEG, electroencephalography.

Beta band analysis also revealed a significant group-by-region interaction effect (χ²=6.28, p=0.0433). In post-hoc tests (Table 3), a significant increase in absolute power was found in the frontal region of the PD group (estimate=3.391, χ²=5.31, p=0.0212), but not in the central (estimate=1.550, χ²=0.86, p=0.3545) or posterior regions (estimate=1.738, χ²=0.93, p=0.3357).

For the alpha band, while a significant group-by-region interaction effect was detected (χ²=6.06, p=0.0483), no significant group differences in absolute power were found in the specific brain regions: frontal (estimate=0.644, χ²=0.02, p=0.9023), central (estimate=-2.202, χ²=0.21, p=0.6483), posterior (estimate=-11.268, χ²=2.13, p=0.1441).

Delta band analysis showed no significant group-by-region interaction effect (χ²=0.80, p=0.6708), but revealed a significant overall group effect (χ²=4.60, p=0.0390), indicating higher delta power in the PD group, though this increase was not region-specific.

Finally, theta band analysis showed no significant group-byregion interaction effect (χ²=0.32, p=0.8520) or overall group effect (χ²=0.35, p=0.5561).

Correlation analysis

Correlation analyses were conducted to examine associations between EEG band powers and clinical symptoms, with all variables visually presented in Figure 2. Given that high beta band power showed the most prominent group difference in the GEE analysis, our analysis focused specifically on the associations between mean high beta power in the frontal, central, and posterior regions and clinical scores on the BDI, BAI, and PDSS.

Figure 2.

Correlation matrix between high beta band power and clinical scale scores in patients with panic disorder. The color scale indicates the strength of Pearson’s correlation coefficients, with blue representing stronger positive correlations and red indicating minimal correlations. A notable positive association was observed between frontal high beta power and PDSS, strong correlations were also observed among the clinical scales themselves. BDI, Beck Depression Inventory; BAI, Beck Anxiety Inventory; PDSS, Panic Disorder Severity Scale.

A significant positive correlation was observed between frontal high beta power and PDSS scores (r=0.41, p=0.010). In addition, strong correlations were found among the clinical scales themselves, including BDI and BAI (r=0.88, p<0.001), BDI and PDSS (r=0.49, p=0.002), and BAI and PDSS (r=0.62, p<0.001). However, no other significant correlations were observed between regional high beta power and the clinical scales.

DISCUSSION

This study explored the electrophysiological mechanisms underlying PD through QEEG analysis during the resting state. Notably, high beta activity was broadly elevated across all regions (frontal, central, and posterior), whereas beta activity showed a more region-specific increase confined to the frontal cortex, consistent with previous research and confirming our hypothesis [17,20,36].

Beta and high-beta oscillations (13–30 Hz) play a critical role in long-range cortical communication, facilitating coordination and efficient information transfer between distributed neural networks [37]. When appropriately tuned, these rhythms help coordinate cognitive processes such as attention, memory, and emotional regulation [26]. However, elevated high-beta activity has a distinct physiological significance. It reflects sustained endogenous workload, using available free energy resources and disrupts the brain’s natural tendency toward synchronization. This keeps the brain in continuous processing away from rest, leading to increased energy consumption and persistent default mode desynchronization [27]. In the present study, elevation of high beta activity suggests that patients with PD maintain a state of anxiety and hyperarousal even at rest, reflecting compromised network efficiency.

The localized frontal beta increase observed in this study also warrants consideration. A temporary increase in beta under stress is considered as a compensatory effort to preserve cognitive control [38]. However, pathologically excessive beta wave activity indicates decreased cognitive flexibility and impaired behavioral control [26]. This pathological profile was evident in PD patients, who showed heightened frontal beta in response to anxiogenic visual stimuli [20]. Our findings suggest that patients with PD exhibit sustained compensatory activation even at rest, which may contribute to their reduced cognitive control and flexibility.

Additionally, beta waves (13–30 Hz) are closely associated with content-specific working memory, reactivating latent memory information as cortical representation within a few hundred milliseconds [39]. Given that patients with PD exhibit heightened threat sensitivity and show conditioned fear responses to stimuli that differ from the original threat [40,41], one possible interpretation is that increased beta range activity (13–30 Hz) reflects repeated reactivation of threat-related representations. Such reactivation could contribute to the persistence of anxious states, even in the absence of immediate external threats.

Notably, fMRI studies have consistently demonstrated amygdala hyperactivation in patients with PD, both during spontaneous panic attacks [42] and in response to panic-related stimuli [43]. In addition, increased prefrontal activation has been observed, particularly during emotion regulation or threat processing tasks [9,43]. This prefrontal activation was thought to reflect emotional stimulus processing and increased cognitive elaboration [9], potentially representing an impaired allocation of prefrontal resources and increased use of maladaptive regulation strategies [43], which are characteristics of anxiety disorders. Our electrophysiological findings of elevated high beta and beta activity are broadly compatible with fMRI observations in PD, showing predominantly frontal topography. This correspondence across neuroimaging modalities supports the evidence for elevated cortical activity patterns in PD.

In addition to the observed changes in beta activity, we investigated alpha activity to deepen our understanding of the neural dynamics in PD. Several studies have reported reductions in alpha activity in PD, particularly in the frontal and temporal regions during resting states [19,23,24]. Alpha oscillations are typically dominant during resting states with minimal attentional demands, reflecting a neural baseline associated with cortical disengagement and low arousal [44,45]. Reduced alpha activity in PD has been interpreted as a shift from this resting-state baseline, potentially indicating heightened cortical arousal or sustained internal alertness [19,23,24]. However, findings across studies remain inconsistent. Some studies have reported increased alpha activity along with elevated delta activity, interpreted as a compensatory mechanism to modulate internal arousal [46]. In our study, a significant group-by-region interaction emerged for alpha activity, yet further analysis revealed no significant regional differences. These findings, along with the heterogeneity of prior results, suggest that the role of alpha activity in PD remains inconclusive and warrants further investigation.

An increase in delta waves (0.5–4 Hz) was observed in this study. Typically associated with deep sleep, delta wave activity has been repeatedly reported in patients with PD. Notably, nocturnal panic attacks have been reported to occur during slow wave sleep (SWS) or during transitions from stage 2 sleep to SWS [47,48]. Additionally, paradoxical delta activity was observed in studies in which panic attacks were pharmacologically induced by sodium lactate in patients with PD [21,22]. Previous research suggests that delta wave activity is associated with primitive brain structures such as the brainstem and plays a pivotal role in immediate responses to threatening situations and in autonomic nervous system regulation. Increased delta activity in PD has been proposed to be linked to activation of structures such as the PAG, responsible for these primitive defense mechanisms [49]. Our findings provide additional clinical evidence supporting this hypothesis.

Overall, our findings suggest that PD is characterized by a state of resting hypervigilance and autonomic dysregulation, supported by increased high beta, beta, and delta activity. These alterations may reflect dysregulated interactions between cortical and subcortical systems, in which excessive cognitive control effort and heightened state of emotional reactivity persist even during rest. In particular, this fast-band profile is compatible with increased cortical activation indicative of sustained hyperarousal at rest, whereas elevated delta activity may reflect activation of autonomic hyperarousal. This electrophysiological profile suggests that aberrant high beta activity, in conjunction with delta changes, may serve as a potential neurophysiological marker of this imbalance in patients with PD.

This study has several limitations that should be acknowledged. As a retrospective study, it lacked randomization and a controlled design, making it difficult to establish clear causal relationships and increases the risk of selection bias. Additionally, the sample sizes were unbalanced between groups, and the overall sample size was relatively small, limiting the generalizability of the findings. Although adjustments were made for age and gender, other confounding variables, such as medication status were not fully standardized, which may have influenced the observed QEEG patterns.

In this study, EEG data were obtained exclusively during the resting state without standardized provocation (e.g., visual anxiety stimuli) and thus do not capture neurophysiological dynamics during actual panic attacks. Due to the limited spatial resolution of QEEG, interpretations regarding specific neuroanatomical sources should be viewed as suggestive.

Given these limitations, future research should address these methodological shortcomings. Larger, adequately powered samples determined by a priori power analyses, together with randomized controlled designs and standardized interventions would reduce bias and improve generalizability. Prospective studies should include structured assessment and reporting of medication exposure.