Attention deficit-hyperactivity disorder (ADHD) is a prevalent neurodevelopmental disorder that is characterized by a consistent pattern of inattention, hyperactivity, and impulsivity [1]. Although ADHD is considered a disabling and common disorder that occurs only in childhood, previous studies have suggested that it persists into adulthood in 30−70% of cases [2-4]. In one previous study, prevalence of adult ADHD in Korea was 71.9 per 100,000, which was increased 10.1 times from 2008 to 2018 [5]. As persistent symptoms of ADHD can have negative impacts on individuals’ lives, families, and society as a whole [6], it is important to precisely recognize it and provide intervention for affected individuals.

The current diagnostic criteria for ADHD based on subjective reports and clinical manifestations are so sensitive that they could not differentiate definite illness from normal variation [1]. In other words, diagnosis can be challenging as core symptoms are nonspecific [7]. From a different perspective, the current diagnostic criteria for ADHD follow the Diagnostic and Statistical Manual of Mental Disorders, fifth edition (DSM-5), which mandates the presence of symptoms prior to the age of 12 years [8]. However, there is a reduction in recall of childhood symptoms over time [9], which can lead to underdiagnosis and overlooking of avoidable negative outcomes and impairments. Thus, there has been a continuous need for biomarkers to reduce misdiagnosis and accurately diagnose ADHD.

Quantitative electroencephalography (QEEG) has been reported as a diagnostic aid for ADHD in many studies [1]. Despite many research studies have been conducted on EEG and childhood ADHD, a significantly smaller number of studies have sought to determine neural correlates and biomarkers in adulthood [10]. Studies have reported consistent elevation of theta (4−8 Hz) levels in adults with ADHD typically associated with focused attention and task difficulty [10]. However, another study [11] has failed to find consistent elevation of theta levels. It also found that theta/beta ratio (TBR) known to be increased in ADHD [12] was not present in ADHD even when multivariate analysis was applied. One of the potential important explanations for such discrepancy in EEG activity in adult ADHD comes from differences in neurological activity according to maturation [13]. To achieve more consistent results, it appears necessary to restrict covariant such as age of study subjects.

In addition to the absolute power of EEG, frontal brain asymmetry has also been studied for its association with ADHD [14-17]. According to a previous study, males in a pediatric sample with ADHD had abnormal alpha asymmetry, specifically showing increased relative right-hemispheric frontal alpha activity known to be associated with approach-related behavior [14]. This pattern of asymmetry was also observed in children with conduct problems [18]. Another study have reported a similar asymmetry pattern in adult ADHD patients, which was observed in resting state EEG compared with a control group [17]. Overall, these studies suggest that individuals with ADHD, particularly males, may exhibit abnormal frontal asymmetry characterized by increased relative right-hemispheric anterior alpha activity, which is associated with approach-related behavior. This pattern of asymmetry appears to be present in both pediatric and adult populations with ADHD, suggesting ontogenetic stability. In conclusion, while QEEG may still hold promise as a potential biomarker for ADHD, given those inconsistent results, further research is needed to better understand its utility. For this reason, young adults aged 18 to 30 years with the highest prevalence of adult ADHD [5] were selected as participants in this study.

This study aimed to determine differences in absolute EEG power and frontal asymmetry between patients with ADHD and non-ADHD in young adults aged 18−30 years. In addition, this study aimed to further explore the possibility of QEEG as a biomarker by performing correlation analysis between QEEG and Stoop test, a neuropsychological test known to evaluate attention [19].

Study Design and Participants

This was a retrospective observational study conducted on patients who first visited the psychiatric clinic of St. Vincent Hospital, the Catholic University of Korea, from January 2018 to June 2020. Participants with psychotic disorders, developmental disease, neurological or severe medical diseases, a history of alcohol or substance abuse/dependence, head trauma, or pregnant women were excluded from this study. The following cases were also excluded: patients who had severe depression (Hamilton Depression Scale ≥ 25) by psychiatric introspection and neuropsychiatry evaluation [20], patients highly suspected of suffering from anxiety disorders through the State-Trait Anxiety Inventory [21], and patients who showed a possibility of bipolar disorder on the Mood Disorder Question-naire [22]. Finally, data from 103 patients (n = 103) were used for analysis. These patients were divided into a group of patients who were later diagnosed with ADHD (n = 51) and age-matched patients without ADHD (n = 52). All patients of non-ADHD group and 30 patients of ADHD group (58.8%) who had comorbid disorder were diagnosed with depressive disorder and anxiety disorder according to the DSM-5 diagnostic criteria [10].

ADHD diagnosis was made by screening with the ADHD Self-Report Scale (ASRS v1.1) [23] upon the first visit to the hospital. Neuropsychological tests were performed for all patients. In particular, the executive function test was used to evaluate patient’s attention [24]. Based on test results, psychiatrists with more than six years of psychiatric clinical practice conducted a final confirmation of ADHD through patient interviews and DSM-5 diagnostic criteria. In the ADHD group, patients were not taking stimulant drugs until EEG was performed. To reduce report bias, EEG data processing and data curation were conducted without knowing which group the subject belonged to. Only one investigator who was conducting statistical analysis with the final diagnosis was unblinded. This study was approved by the Institutional Review Board of St. Vincent’s Hospital, The Catholic University of Korea (approval number: VC20RISI0128).

Neurocognitive Function Test

A neurocognitive function test was performed to compare general cognitive functions of the two groups. General cognitive functioning was measured using the Korean Wechsler Adult Intelligence Scale-IV (K-WAIS-IV), the most frequently used and validated measure of the intelligence quotient (IQ) in Korean individuals aged 16 to 69 years. The K-WAIS-IV includes 15 subtests comprising the Full-scale IQ (FSIQ) and four indices: the Verbal Comprehension Index (VCI), Perceptual Reasoning Index (PRI), Working Memory Index (WMI), and Processing Speed Index (PSI) [25]. The FSIQ score reflects the overall level of an individual’s intellectual functioning. The VCI measures verbal ability including verbal comprehension, reasoning, and conceptualization. The PRI assesses abilities in nonverbal conceptual reasoning, visual organization, and perception. The WMI reflects the ability to concentrate and working memory capacity. The PSI represents psychomotor speed in response to nonverbal, visual stimuli [26].

Neuropsychological tests [24] including the Stroop test were also used to assess executive function. The Stroop test is well known as a measure of selective attention and cognitive control [19]. This study used the Korean version of the Stroop test [24]. It consists of three trials. In the first trial, called the “simple task,” patients should say names of colors of dots printed in colored ink (e.g., red, blue, black, yellow). In the second trail, called the “middle task,” participants must read words (e.g., when, go, come) printed in color. In the final trial, called the “interference task,” the subject must say the word’s color instead of the word itself, which is the name of a color other than the one in which it is printed. This test is scored according to response time. The standard score for the simple task and that of the interference task were used in our analysis. The standard score was calculated as a normal distribution score with a mean of 10 and a standard deviation of 3 based on previous study [27].

EEG Procedure and Processing

EEG was recorded on a Nihon Kohden digital EEG device (EEG 2100) at the following 19 electrodes placed on the scalp based on the International 10/20 system: Fp1, Fp2, F3, F4, F7, F8, Fz, T3, T4, C3, C4, Cz, T5, T6, P3, P4, Pz, O1, and O2. Two additional electrodes (A1, the ground electrode; and A2, the reference electrode) were placed on ear lobes. Electrocardiograms were recorded on separate channels. All data were processed with a 0.1−100 Hz band pass filter and sampled at 500 Hz. All EEG acquisitions were conducted by one technician in an electrically shielded EEG room in the Department of Neurology at St. Vincent Hospital. Participants sat on a comfortable chair with eyes closed. They were instructed to remain awake during EEG acquisition. EEG recording lasted for 20 minutes.

For EEG data pre-processing, 60 non-overlapping EEG screen segments of 5 seconds were selected in each situation by visual inspection. Gross artifacts such as artifacts caused by movement were rejected through visual inspection by a trained person without prior information regarding origin of data. Epochs with signals exceeding ± 90 mV on any channel were rejected from the analysis. To remove artifact, independent component analysis was performed using EEGLAB toolbox and MATLAB R2023a (MathWorks) [28].

After removing artifact, the acceptable width of the EEG data was smoothed using Fast Fourier transformations (FFT) and averaged over four frequency bands with a NeuroGuide’s spectroscopic system at the absolute (mV) power. After performing FFT, the spectral density was averaged over a special frequency range. Frequency band ranges represented delta (1−4 Hz), theta (4−8 Hz), alpha (8−12 Hz), beta (12−25 Hz), high beta (25−30 Hz), and gamma (30−40 Hz) frequencies. We also calculated the ratio between theta and beta absolute power in order to obtain TBR as described in previous studies [7]. Since abnormalities in resting QEEG associated with ADHD symptoms such as impulsivity and inattention were commonly detected in frontal regions [29-31], we focused on the frontal region when examining resting QEEG measures. In this study, we divided electrodes of the frontal lobe into three parts, left frontal (Fp1 + F3 + F7), mid frontal (Fz), right frontal (Fp2 + F4 + F8), as reported previously [32]. Secondly, frontal asymmetry was analyzed for each frequency. Electrodes were divided into three paired groups of prefrontal (Fp2-Fp1), middle frontal (F4-F3), and lateral frontal (F8-F7) based on a previous study [33].

Statistical Analysis

All parameters were divided into two groups: patients with ADHD and those without ADHD (non-ADHD). To compare differences in demographic data and clinical measurements, both groups were compared using the chi-square test for discontinuous variables. For continuous variables, after verifying whether normality assumption was satisfied by the Shapiro–Wilk test, Mann–Whitney Utest or independent t test was used for comparison.

In EEG analysis, absolute powers of patients were first evaluated using repeated measures analysis of variance (ANOVA) with electrodes (Fp1, Fp2, F3, F4, F7, F8, Fz) and waves (delta, theta, alpha, beta, gamma) as within-subject factors and groups (ADHD vs. non-ADHD) as between-subjects factors. Covariates included age, sex, and FSIQ. To analyze asymmetry, the frontal asymmetry index (FAI) was defined using the following standard calculation: [(R − L)/(R + L) × 100], which was used in a previous study [33]. A positive asymmetric index yields a dominant brain activity at the right hemisphere, while a negative index yields a dominant brain activity at the left hemisphere. To find differences between the two groups, repeated measures ANOVA also was used with age, sex, and FSIQ as covariants. Additionally, as there was a disparate distribution of comorbidities between the two groups, comorbidity was included as a covariate to modify this discrepancy.

If the ANOVA revealed a significant interaction effect with group, multiple covariate ANOVA was performed for each frequency to analyze results in more detail. At the same time, based on the existing literature that revealed correlation between reaction time of the stroop test and neural activation of EEG [34,35], partial correlation analysis was conducted between the standard score of Stroop tests and absolute powers, FAI of both ADHD group and non-ADHD group, covariates were age, sex, FSIQ and comorbidity. Differences were considered significant at p ≤ 0.05. All statistical analyses were performed using SPSS version 26.0 (SPSS Inc.).

Demographic and Clinical Characteristics

Data from a total of 103 patients were used for the analyses, of which 51 were diagnosed with ADHD and 52 had non ADHD. Table 1A presents demographic data, clinical measurements, and comorbidity of all patient groups. There was no significant difference in age (p = 0.650) or sex (p = 0.118) between the two groups. In terms of comorbidity, 30 patients with ADHD (58.8%) had comorbidity and all patients of non-ADHD had cormobid psychiatric disease. In details, 24 patients with ADHD (47.1%) had depressive disorder and 6 (11.8%) had anxiety disorders, while 35 patients with non-ADHD (67.3%) had depressive disorder and 16 (32.7%) had anxiety disorder. On the other hand, non-ADHD patients had a significantly higher proportion of those with a diagnosis of depression than ADHD patients (χ² = 28.305, p < 0.001). For clinical characteristics, there was no significant difference except for ASRS (p < 0.001), with ADHD patients having a significantly higher ASRS score than non-ADHD patients. The Hamilton Depression Scale and the State-Trait Anxiety Inventory did not show significant differences between the two groups.

Additionally, a subset of 24 patients (4 ADHD [7.84%], 20 non-ADHD [38.5%]) were on medication, with 3.92% of those in the ADHD group and 28.8% of those in the non-ADHD group on antidepressants, 0% of those in the ADHD group and 3.85% of those in the non-ADHD group on atypical antipsychotics, and 5.88% of those in the ADHD group and 26.9% of those in the non-ADHD group on anxiolytics (Table 1B). Differences in medication usage were not statistically significant (antidepressants: p = 0.414; atypical antipsychotics: p = 0.178; anxiolytics: p = 0.593). Also in Table 2, neurocognitive function test, there was no significant differences between two groups.

EEG Absolute Power

In the comparison for the absolute power, there was a significant difference between the two groups (Table 3). As expected, there was a main effect of between groups (F (1, 98) = 5.021, p ≤ 0.05). To understand which regions and which frequencies had significant differences, multivariant analysis of covariance (MANCOVA) was conducted for post hoc analysis (Table 3). First in delta band, the ADHD group showed significantly more delta power in the mid frontal region than the non-ADHD group (p = 0.035). In beta band, the middle frontal region was lower (p = 0.021) in the ADHD group. The high beta band also had significant difference between the two groups (Lt. [p = 0.021], Mid. [p = 0.017], Rt. [p = 0.046]). There was no significant difference in theta, gamma band, or TBR between the two groups (Table 3).

FAI

In the comparison for FAI, there was also a significant difference between the two groups (Table 4). First, there was a main effect of between groups (F (1, 98) = 6.315, p = 0.014). MANCOVA revealed that there were significant differences at all frequencies. In details, middle frontal (delta: p < 0.001; theta: p = 0.001; alpha: p = 0.014), lateral frontal (delta: p = 0.003; alpha: p = 0.023) in low frequencies (delta, theta, alpha), and prefrontal (beta: p = 0.016; high beta: p = 0.030), and middle frontal (beta: p = 0.004; high beta: p = 0.018; gamma: p = 0.019) were significantly different in high frequencies (beta, high beta, gamma).

Correlations

We selected electrodes and frequencies which had significant differences between the two groups after repeated measures ANOVA and MANCOVA. After selection, partial correlation was performed to find correlations between the standard score of both reaction time (RT) of Stroop test and EEG of both groups after modifying covariant (Table 5). In Table 5, the score of simple task RT was negatively correlated with mid frontal delta wave (r = −0.260, p = 0.009) but positively correlated with mid frontal beta wave (r = 0.227, p = 0.023). Also, the standard score of interference RT was negatively correlated with mid frontal delta (r = −0.237, p = 0.017) but positively correlated with mid frontal beta (r = 0.227, p = 0.025) waves.

Partial correlation analysis was also performed between FAI and Stroop tests. Results revealed that middle frontal asymmetry at all frequencies were negatively correlated with both standard score of RT (Table 6).

This study aimed to explore differences in QEEG between young adults aged 18−30 years who were diagnosed with ADHD and those without ADHD. In this study, we found significantly different EEG absolute powers at both frontal areas in some frequencies and FAI in each frequency. Furthermore, we found that differences in absolute power and FAI between the two groups showed significant correlations with the score of the Stroop test, a neuropsychological test known to reflect attention. Unexpectedly, there was no significant difference in neuropsychological test between the two groups.

First in EEG, the ADHD group had higher absolute power in delta band but lower absolute power in beta band at the middle frontal region. This finding is consistent with results of previous studies showing that increases of delta and theta waves were commonly observed in ADHD with a high level of reliability and that the delta wave was increased mainly in the frontal lobe [36]. More recently, researchers have also found that adult patients with ADHD and their first-degree relatives show increased delta power in frontal region than non-ADHD controls in eye-closed EEG spectral analysis [37]. In contrast, decreased absolute powers of alpha and beta waves were observed in individuals with ADHD [36,38]. These findings were corroborated by a functional imaging study [39], which provided an explanation for the association between frontal lobe hypofunction and deficits in attention and behavior control. Our study also found that the delta wave was negatively correlated with the Stroop test score and that the beta wave was positively correlated with the Stroop test score. Considering that the Stroop test reflects participants attention and response inhibition [40,41], increased delta and decreased beta wave in adult ADHD might be related to impairment of controlling attention and behavior.

In frontal asymmetry, we also found that there were significant differences in the middle frontal region (F4-F3) at all frequencies between the two groups. Frontal asymmetry at all frequencies in middle frontal region were negatively correlated with both scores of the Stroop test. The ADHD group had higher FAI than the non-ADHD group, meaning more rightward activated frontal function. In previous studies, individuals with ADHD exhibited greater rightward frontal alpha activation compared to non-ADHD individuals in both children [14] and adults [15-17]. Previous studies have proposed that these findings are associated with the approach-withdrawal model of ADHD [15,42]. They found that ADHD patients had motivational deficit, which could make symptoms worse. Alpha asymmetry could reflect the deficit of motivation in adults with ADHD. Frontal alpha asymmetry which was correlated negatively with the Stroop test in our study could support these hypotheses. Other researchers have also found correlations of functional magnetic resonance imaging and EEG theta with alpha asymmetry [43]. Regarding beta band, previous studies have also revealed a rightward asymmetry in adult ADHD patients, although it is limited to the temporal-parietal area [44,45]. One study has also found correlations between increased right frontal gamma wave, not asymmetry, and ADHD symptoms with mood disorders [31]. Integrating results of these various studies, frontal asymmetry in all frequencies might have potential as a biomarker of ADHD. This should be studied in the future.

TBR showed no significant difference in frontal area of our study, consistent with results of a systematic review [10]. Previous studies suggested that brain maturation could affect the TBR, which could explain why childhood ADHD patients showed significant difference of TBR, but not adult ADHD patients [10,46].

To the best of our knowledge, there are not many studies on asymmetry in delta bands. However, in one study, negative cognition and mood were associated with rightward delta and theta wave in post-traumatic stress disorder patients compared to the control group, with life-time stress events and hyperarousal associated with rightward theta, alpha, and beta waves [47]. Previous studies have found that childhood trauma experiences are related to adult ADHD [48-50]. Thus, asymmetry in these various frequencies might also have the potential as a biomarker for adult ADHD. This merits further research.

Although we expected that the ADHD group would have higher z-score of the Stroop test RT, there was no statistically significant differences in the Stroop test score despite the fact that the ADHD group had lower score than the non-ADHD group. There are several points of view regarding this result. First, ADHD patients were able to compensate deficient activation of attention-related neural generators. One study has found that there are significant differences in ADHD patients’ EEG while N-back working memory test shows no significant differences [51]. It has been suggested that this result could support neural compensatory mechanism hypothesis [51]. Second, the Stroop test is less sensitive tool to reflect the intension of adult ADHD than other neuropsychiatric tests. Two meta-analyses [52,53] have found that the Stroop interference effect, which reflects response inhibition, is not larger in individuals with ADHD than in the control group. Additionally, maturation rates of both ADHD individuals and controls were similar, indicating no differential development in terms of Stroop performance [52]. Third, there was a heterogeneity of patients with ADHD in this study. In our study, we did not classify patients with ADHD into combined type, inattention type, or hyperactivity type. One study has found that patients with ADHD, predominantly of the hyperactive type, show no deficits in Stroop test or GO/No-Go task [54]. Another study [55] has found no significant correlation between Continuous Performance Test (CPT) and the Stroop test. That study explained that findings of differentiation in performance in different tests could support the role of cognitive domains in the subtyping of ADHD. Neuroimaging studies [56,57] have also found partial overlapped activation of brain areas while performing the CPT or the Stroop test, supporting this hypothesis. Therefore, in future studies, it is necessary to classify ADHD patients to subtypes and perform tests related to ADHD symptoms such as CPT. Despite the above limitations of Stroop tests, differences of delta and beta waves in absolute powers of EEG between the two groups showed significant correlations with Stroop test results. Based on these results, we could interpret that EEG measures including frontal asymmetry might reflect frontal lobe function and potentially be more sensitive indicators of differences between individuals with ADHD and those without than the Stroop test alone. This suggests that EEG could provide additional insights into the underlying neural mechanisms associated with ADHD and may capture aspects of attention and cognitive control not fully captured by the Stroop test.

This study has some limitations. First, we could not fully control the effect of medication on the EEG of patients. As previously known, many psychotropic drugs can affect the EEG [58]. We could not fully correct various effects of drug use on EEG. Thus, future studies need to study drug-naive patients. Second, ADHD groups was not classified according to the detailed subtype of the ADHD. Since differences of EEG according to the heterogeneity of ADHD were not controlled, EEG changes according to the subtype of ADHD should be considered in future studies. Third, in this study, clinical characteristics and all components of the EEG in ADHD and non-ADHD patients were not compared with normal control groups. Finally, since it was cross-sectional in nature, a longitudinal study is needed to further evaluate dynamic changes in the human brain in the future.

Nevertheless, this study was an important attempt to compare absolute powers and frontal asymmetry of QEEG between psychiatric patients with ADHD and those without in young adults. Moreover, it also evaluated the correlation between neuropsychological findings known to reflect attentional deficit, behavioral deficit, and EEG findings. Future studies should explore various aspects of attention and behavioral control in adults with ADHD.

No potential conflict of interest relevant to this article was reported.

Conceptualization: Sung-Hoon Yoon, Jihye Oh and Jong-Hyun Jeong. Data acquisition: Sung-Hoon Yoon, Jihye Oh, Jong-Hyun Jeong, Yoo Hyun Um, Ho Jun Seo, Tae Won Kim and Seung Chul Hong. Supervision: Jong-Hyun Jeong. Writing—original draft: Sung-Hoon Yoon. Writing—review & editing: Jong-Hyun Jeong.