Neuropsychology Copyright 1999 by the American Psychological Association, Inc.

1999, Vol. 13, No. 3, 424-433 0894-4105/99/S3.00

Assessing Attention Deficit Hyperactivity Disorder via Quantitative

Electroencephalography: An Initial Validation Study
Vincent J. Monastra Joel F. Lubar
The Family Psychology Institute University of Tennessee

Michael Linden Peter VanDeusen

Mission Psychological Consultants Attention Development Programs

George Green William Wing

The Biofeedback Center Attention Deficit Treatment Center of Cincinnati

Arthur Phillips T. Nick Fenger

La Jolla, California St. Louis Psychological and Educational Associates

Spectral analysis of the electrophysiological output at a single, midline prefrontal location (the
vertex) was conducted in 482 individuals, ages 6-30 years old, to test the hypothesis that
cortical slowing in the prefrontal region can serve as a basis for differentiating patients with
attention deficit hyperactivity disorder (ADHD) from nonclinical control groups. Participants
were classified into 3 groups (ADHD, inattentive; ADHD, combined; and control) on the basis
of the results of a standardized clinical interview, behavioral rating scales, and a continuous
performance test. Quantitative electroencephalographic (QEEG) findings indicated significant
maturational effects in cortical arousal in the prefrontal cortex as well as evidence of cortical
slowing in both ADHD groups, regardless of age or sex. Sensitivity of the QEEG-derived
attentional index was 86%; specificity was 98%. These findings constituted a positive initial
test of a QEEG-based neurometric test for use in the assessment of ADHD.

Attention deficit hyperactivity disorder (ADHD) is a Biederman et al. (1990), Mannuzza and colleagues (Man-
psychiatric disorder that has been historically characterized nuzza, Klein, Bessler, Malloy, & LaPadula, 1993; Mannuzza
by the behavioral symptoms of inattention, impulsivity, and et al., 1991), and Morrison (1980).
hyperactivity (American Psychiatric Association, 1980,1987, Consistent with the behavioral formulation of ADHD,
1994). Estimates of the prevalence of this disorder range psychometric procedures initially focused on assessing this
from 5% to 15% of the school-age population (American disorder from a behavioral perspective. Behavioral rating
Psychiatric Association, 1994; Barkley, 1990; Rie & Rie, scales, such as the Child Behavior Checklist (Achenbach &
1980), occurring more commonly in boys than girls (ratios Edelbrock, 1983), the Conners' rating scales (Conners,
range from 4:1 to 9:1). Onset typically occurs prior to age 7. 1973), the ADHD Rating Scale (DuPaul, 1991), and the
The condition is a relatively enduring one (Barkley, 1997a), Attention Deficit Disorder Evaluation Scale (McCarney,
and children diagnosed with ADHD are at increased risk for 1989), were developed and provided a database for compar-
emergence of comorbid psychiatric disorders, as reported by ing the behavioral observations of parents and teachers with
normative populations. Similarly, performance tests measur-
Vincent J. Monastra, Clinical Director, The Family Psychology
ing capacity for vigilance and impulse control during visual
Institute, Endicott, New York; Joel F. Lubar, Department of and auditory tracking tasks (continuous performance tests)
Psychology, University of Tennessee; Michael Linden, Mission were developed (Conners, 1994; Gordon, 1983; Greenberg,
Psychological Consultants, San Juan Capistrano, California; Peter 1994; Sanford, 1994), providing a more objective measure
VanDeusen, Attention Development Programs, Atlanta, Georgia; of the core symptoms of inattention and impulsivity. As
George Green, The Biofeedback Center, Reno, Nevada; William Barkley (1990), Trommer, Hoeppner, Lorber, & Armstrong
Wing, Attention Deficit Treatment Center of Cincinnati, Cincinnati, (1988), and Cantwell (1996) concluded, these measures are
Ohio; Arthur Phillips, Independent Practice, La Jolla, California; T. useful in the assessment process, particularly when com-
Nick Fenger, St. Louis Psychological and Educational Associates, bined with a thorough review of medical, developmental and
St. Louis, Missouri.
We thank Donna Monastra for her review of the initial draft of
family histories and an examination of intellectual functions
this article and Diane Taylor and Michelle Coins for their and academic achievement. However, as Barkley, Trommer
assistance in data entry and preparation of the manuscript. et al., and Cantwell noted, these tests cannot be considered
Correspondence concerning this article should be addressed to diagnostic for ADHD because of the rater bias associated
Vincent J. Monastra, Clinical Director, The Family Psychology with rating scales and the high false-negative rate reported
Institute, 2102 East Main Street, Endicott, New York 13760. with continuous performance tests. In order to improve

424
 ASSESSING ADHD VIA QEEG 425

diagnostic accuracy, the development of additional assess- the EEG protocols of ADHD versus control groups. Mann et
ment procedures appeared necessary. al. (1992) showed significant increases in slow-wave activ-
As reviewed by Barkley (1997b), the primary deficits ity (4.00 to 7.75 Hz) in prefrontal, midline regions, with
associated with ADHD support a hypothesis that anatomical decreased posterior beta activity (12.75 to 21.00 Hz) when
and biochemical abnormalities of the prefrontal cortex EEG recordings were obtained during academic challenges.
constitute the physical basis of this disorder. Physical Similarly, Janzen, Graap, Stephanson, Marshall, and Fitzsim-
examination of this cortical region has been conducted with mons (1995) noted increased theta activity in frontal,
neurodiagnostic procedures (e.g., positron emission tomog- central, and posterior regions. Lubar (1995; Lubar, Swart-
raphy [PET] and single photon emission tomography wood, Swartwood, & Timmermann, 1996) examined the
[SPECT]). The results of these examinations have provided relationship between ADHD and a ratio derived by dividing
evidence of hypoperfusion and low metabolic activity in the the electrophysiological output (pW) produced in a fre-
prefrontal and caudate nuclei regions (Amen, Paldi, & quency band defined as 4 to 8 Hz by the output produced in
Thisted, 1993; Lou, Henriksen, & Bruhn, 1984; Zametkin et frequencies from 13 to 21 Hz. This theta-beta power ratio
al., 1990; Zametkin & Rapoport, 1987). In addition, neuro- was calculated as individuals completed the following tasks:
imaging procedures (e.g., magnetic resonance imaging eyes open baseline, eyes closed baseline, reading silently,
[MRI]) have revealed anatomical differences in the caudate completing visuomotor tasks, and listening. Lubar and his
nucleus (Casey et al., 1997; Hynd et al., 1993) and corpus colleagues hypothesized that evidence of excessive cortical
callosum (Hynd et al., 1991). Overall, as noted by Casey et slowing (i.e., a higher ratio of slow-wave activity relative to
al. (1997), these studies have provided clear evidence of the fast EEG activity) would be noted in individuals with
importance of the frontostriatal circuitry (specifically, in the ADHD. Their findings supported this hypothesis. Significant
right hemisphere) in understanding the neurological basis of group differences were noted in the theta-beta power ratios
ADHD. obtained at multiple cortical sites, with CZ and FZ appearing
Three types of research initiatives, stimulated by the the most promising for consideration in the development of
results of these neurological studies, emerged in an effort to an assessment procedure on the basis of spectral analysis.
improve diagnostic accuracy. Each research initiative exam- Chabot and Serfontein (1996) expanded this research in
ined procedures that assess the functional performance or their examination of 310 "normal" and 407 attention deficit
electrophysiological activity of the frontal lobes. These disorder-attention deficit hyperactivity disorder patients.
research efforts have included neuropsychological studies Initially, by using a discriminant function analysis of
assessing the performance of individuals with ADHD on multiple QEEG characteristics, they correctly identified
tests associated with frontal lobe functions (reviewed by approximately 95% of the normal and 93% of the ADD-
Barkley, Grodzinsky, & DuPaul, 1992), quantitative electro- ADHD patients. In their subsequent study, Chabot et al.
encephalographic (QEEG) studies examining event related (1996) sought to examine the sensitivity and specificity of
potentials in individuals with ADHD (e.g., Kuperman, their procedure in an examination of 407 children with
Johnson, Arndt, Lindgren, & Wolraich, 1996; Linden, Gev- attentional disorders and 242 children with learning disor-
irtz, Isenhart, & Fisher, 1996; Loiselle, Stamm, Maitinisky, ders. Similar to their earlier findings, Chabot et al. (1996)
& Whipple, 1980; Satterfield, Schell, Nicholas, Satterfield, reported 93% correct classification of the children with
& Freese, 1990), and QEEG studies using computerized ADHD and 90% of children with learning disorders when a
power spectral analysis (PSA) to study patterns of cortical discriminant function analysis of nine QEEG measures was
activation (e.g., Capute, Niedermeyer, & Richardson, 1968; conducted.
Chabot, Merkin, Wood, Davenport, & Serfontein, 1996; On the basis of the previous QEEG studies that used
Klinkerfuss, Lange, Weinberg, & O'Leary, 1965; Lubar, power spectral analysis, our research team sought to develop
1991; Mann, Lubar, Zimmerman, Miller, & Muenchen, and test a simplified neurometric procedure for use in the
1992). assessment of ADHD. Prior findings (Lubar, 1995; Lubar et
The present study proceeded from an examination of the al., 1996) have indicated that the highest degree of differen-
QEEG studies conducted using PSA. Such procedures tiation between ADHD and non-ADHD participants was
involve the collection of multiple, short periods of digitized noted at the vertex; thus, CZ was selected for placement of
electroencephalographic (EEG), which are subjected to a the active electrode. Because critiques of prior QEEG
fast Fourier transformation (FFT) algorithm (Cooley & studies (Levy, 1994) noted that the statistical differences
Tukey, 1965). The FFT-derived data are then averaged over between groups could have occurred as a function of
all trials for a given experimental condition. The overall multiple statistical comparisons, only one active site was
electrophysiological power (pW) can then be determined used. Previous studies have indicated differentiation be-
and compared for various frequency bands at each active tween groups when participants were involved in scholastic
electrode site. Common frequency bands investigated by tasks (e.g., reading, listening, drawing), and difficulty sustain-
researchers have included delta (0.1 to 4 Hz), theta (4 to 8 ing attention during completion of these types of tasks
Hz), alpha (8 to 12 Hz), sensorimotor rhythm (12 to 16 Hz), frequently results in referral of children for evaluation;
and beta (16 to 20 Hz). therefore, QEEG recordings were obtained while children
Systematic, multisite spectral analysis studies comparing completed reading, listening, and drawing tasks.
QEEG data of patients with ADHD and nonclinical controls In order to minimize experimenter bias, evaluations were
have revealed certain cortical locations that differentiated conducted by members of our research team at eight
426 MONASTRA ET AL.

independent locations. To reduce error due to low interrater groups free of other neurological conditions, caretakers completed
reliability rates for ADHD (reviewed by Barkley, 1990), Barkley's (1991a) ADHD Clinical Parent Interview for participants
classification as ADHD or non-ADHD was based on data aged 6 to 20 years. Participants aged 21 to 30 years completed the
derived from a combination of a structured clinical inter- adult version of this structured interview. Detailed information
regarding medical and developmental history was obtained through
view, a behavioral rating scale, and a continuous perfor-
this interview. Individuals with other neurological disorders (e.g.,
mance test. Because the most commonly used behavioral epilepsy, autism) were not included in this study. All participants
rating scales provide standard scores for inattentive or were under the care of physicians. None reported treatment for any
combined hyperactive-impulsive symptoms, only two of the neurological condition.
ADHD subtypes (i.e., ADHD, inattentive, and ADHD, To control for medication effects, none of the members of the
combined) were examined. Predominately hyperactive or control group were evaluated while using any medication. For
impulsive types of ADHD patients who did not test positive those participants being treated with Ritalin, testing was completed
for inattention were not examined in this study because of after a medication-free period of at least 12 hr. Given the clinical
our effort to reduce classification error. A continuous perfor- action of this medication as well as published research (Lubar et al.,
mance test was added to the screening process used by Mann 1996) that has indicated no effect of stimulant medication on the
QEEG recordings obtained from 19 sites (including CZ), we
et al. (1992) and Chabot et al. (1996) because such
considered our clinical groups to be medication free as well.
procedures have been shown to reflect frontal lobe function- Evaluations were conducted between the hours of 9:00 a.m. and
ing by SPECT (Rezai et al., 1993), are useful in identifying 3:00 p.m.
individuals with attentional deficits, and have been associ- Classification of individuals into clinical and nonclinical groups
ated with low false-positive rates for ADHD (Greenberg, was accomplished through a screening procedure that included
1994). Barkley's ADHD Clinical Parent Interview (or Adult Interview;
Given the findings of previous QEEG studies, we hypoth- Barkley, 1991a), behavioral rating scales (Attention Deficit Disor-
esized that significant differences in the theta-beta power ders Evaluation Scale [ADDES; McCarney, 1989], ADD-H: Com-
ratios would be noted, with patients diagnosed with ADHD prehensive Teacher's Rating Scale [Ulmann, Sleator, & Sprague,
exhibiting higher ratios than nonclinical controls. In order to 1984], or other ADHD rating scales), and a continuous perfor-
mance test (Conners' Continuous Performance Test [Conners,
initially test the classification accuracy of a neurometric test
1994], Gordon Diagnostic System [Gordon, 1983], Test of Vari-
based on the theta-beta power ratio, critical values of the ables of Attention Continuous Performance Test [Greenberg,
power ratio were to be calculated on the basis of the mean 1994], and Intermediate Visual and Auditory Continuous Perfor-
and standard deviation of the control groups. It was hypoth- mance Test [Sanford, 1994]). To be placed in one of the clinical
esized that classification of participants into ADHD and groups, participants had to meet the Diagnostic and Statistical
non-ADHD groups could be made on the basis of these Manual of Mental Disorders (4th ed.; DSM-IV; American Psychiat-
critical values, given the location of our QEEG recording ric Association, 1994) criteria for ADHD on the basis of the report
site and the neuroanatomical and biochemical research data of the referral source and had to test positive for ADHD on both
supporting the role of the prefrontal cortex in ADHD. Our behavioral and continuous performance test measures. The specific
goal was to conduct the initial validation study involving a criteria for group placement were as follows: For ADHD, inatten-
tive type (ADHD-I), participants had to meet DSM-IV criteria
specific neurometric indicator of cortical slowing, in order to
according to a referring source (school, physician), meet DSM-IV
begin the process of developing an inexpensive, nonintru- criteria according to caretaker or self-report on the Barkley
sive electrophysiological measure of frontal lobe function- Interview, obtain a positive score for inattention on the ADDES or
ing that could contribute to the existing assessment proce- other ADHD rating scale and score in the nonclinical range on the
dures for the diagnosis of ADHD. Impulsive and Hyperactive scales, and obtain a positive overall
rating for ADHD on a continuous performance test. For ADHD,
Method hyperactive-combined type (ADHD-H/C), participants had to meet
the same first two requirements as for ADHD, inattentive type, to
Participants obtain a positive score for impulsivity or hyperactivity on the
ADDES or other rating scale, and to obtain a positive overall rating
Four hundred and eighty-two individuals were evaluated using for ADHD on a continuous performance test. Finally, for nonclini-
behavioral rating scales, continuous performance tests, computer- cal controls, participants did not meet DSM-IV criteria for any
ized PSA of QEEG recordings, and structured clinical interviews. psychiatric disorder on the basis of caretaker or self-report on the
Two hundred and seven of the participants were girls, female Barkley Interview, their caretaker or self-report scores on the
adolescents, and women; 275 were boys, male adolescents, and ADDES or other ADHD rating scale were in the nonclinical range
men. In an effort to minimize experimenter bias and obtain data for inattention, impulsivity, and hyperactivity, and their continuous
from multiple geographic regions, eight research centers in the performance test overall ratings were negative for ADHD. Distribu-
following states participated in the project: New York, Georgia, tion of participants by age and diagnosis is presented in Table 1.
Ohio, Tennessee, Missouri, Nevada, and California. The geo-
graphic distribution of participants was as follows: Eastern Region
(New York) = 24%, Central Region (Ohio, Tennessee, Missouri) = Materials
29%, Southern Region (Georgia) = 36%, and Western Region
(California, Nevada) = 11%. Participants were recruited by QEEG recordings were obtained using Autogenics A-620 Elec-
correspondence with schools, physicians, and mental health profes- troencephalograph (Wood Dale, IL) with associated Assessment
sionals located near each of the participating research centers, as Software (Wood Dale, IL) for computerized analysis of EEG data.
well as through newspaper solicitation. This system provides researchers with a quantitative analysis of
Because of the importance of establishing clinical and control electrophysiological recordings in multiple frequency bands. For
 ASSESSING ADHD VIA QEEG 427

Table 1 and read by the clinician (as described for the reading task). EEG
Distribution of Participants by Age and Diagnosis review was conducted as with the first two tasks. The final task was
drawing. A stable drawing surface was placed in front of the child
Age (years) or adult. He or she was instructed to copy geometric figures from
Diagnosis 6-11 12-15 16-20 21-30 one of the following tests: Beery Developmental Test of Visual-
Motor Integration (Beery & Buktenica, 1967), Benton Visual
ADHD-I 64 48 51 13 Retention (Benton, 1955), or McCarthy Scales of Children's
ADHD-H/C 149 43 21 8
Control 30 34 10 11 Abilities (McCarthy, 1972). EEG was recorded for 90 s, with re-
Total 243 125 82 32 view as with the previous tasks.
Note. ADHD-I = attention deficit hyperactivity disorder, inatten-
tive; ADHD-H/C = attention deficit hyperactivity disorder, hyper- Results
active-combined type.
Cortical Slowing and ADHD
the purpose of this study, 4—8 Hz defined the theta band, and 13-21 The initial statistical analyses were conducted in order to
Hz defined the beta band. Similar to other PSA studies, multiple test the hypothesis that participants identified with ADHD
short periods (90 s) of digitized EEG were obtained. An FFT (either inattentive or combined types) would display signifi-
algorithm was computed by the A-620 Assessment System and
cantly higher levels of slow-wave (i.e., theta, 4-8 Hz)
averaged over four trials. The overall electrophysiological power
(pW) was computed for the theta and beta bands by the A-620 relative to fast-wave EEG activity (i.e., beta, 13-21 Hz). The
Assessment System and then manually entered into the statistical calculation of these theta-beta power ratios was performed
program Statistica (StatSoft, 1995) for data analysis and graphic by the A-620 Assessment Software for each participant on
presentation. each of four tasks. The resulting ratio data was then
transferred to StatSoft's Statistica program for statistical
Procedure analysis and graphic presentation of data.
The planned statistical analysis consisted of an analysis of
Participants meeting selection criteria for involvement in this variance (ANOVA) with repeated measurement of the
study were evaluated using the following QEEG procedure: theta-beta ratio during four tasks (baseline, reading, listen-
1. The vertex (CZ) was located using the International 10-20 ing, and drawing). Between-subject comparisons were made
System of electrode placement (Andreassi, 1989). to examine the effects of age and diagnosis on the theta-beta
2. The area was cleaned using Omni prep (or equivalent) and power ratio. Within-subject comparisons were studied in
isopropyl alcohol. A small amount of conductive paste (e.g., Ten20) order to evaluate task effects. Tukey's honest significant
was applied to the scalp and to a Grass Gold Disc Electrode
(Astro-Med, Inc., West Warwick, RI) with hole (E5GH), and the difference (HSD) test was selected for post hoc testing of
sensor was attached to the scalp. A similar cleaning procedure was significant main or interactional effects. An alpha level of
used for preparing the earlobes and one pair of Gold Disc .01 was used for all statistical tests.
Electrodes in Ear Clip (Grass E34D) was attached to each earlobe. A summary of the ANOVA analysis of all effects is
Quality of preparation was assessed by way of an Autogenics provided in Table 2. Consistent with our hypothesis, statisti-
Electrode Tester (Wood Dale, IL). Impedance readings were to be cal analysis revealed that theta-beta power ratios were
below lOKfl. Offset potential was to be below 10 uV before significantly affected by age and diagnosis (p < .001). In
recordings were obtained. addition, data analysis indicated that the power ratio was
3. Band frequencies were defined on the Assessment Software affected by type of task (p < .001). There was no evidence
with 4-8 Hz defining theta and 13-21 Hz defining beta. Once the that the degree of cortical slowing was related to the sex of
sensors were tested and band frequencies defined, the participant's
EEG activity at CZ was recorded during four tasks. The first task the participant (Rao's R = .646, p = .63). Similarly, there
was eyes fixed-baseline. The child or adult was seated in front of was no indication that the effects of age, diagnosis, or task
the computer monitor display and instructed to focus his or her were confounded by the sex of the participant.
gaze on the monitor's "on/off' indicator light. EEG recordings Post hoc comparisons of the main effects (age and
were obtained for 90 s. After the task was completed, the EEG diagnosis) consisted of examination of the theta-beta power
record was reviewed in 2-s intervals (epochs), in order to manually ratios on each of the four tasks. Consistent with maturational
filter out epochs containing excessive electromyograph (EMG) models of cortical development, the level of cortical slowing
artifact (e.g., body movement, eye rolls or blinks). A minimum of noted in our PSA study was highest in the youngest age
15 low-artifact epochs (i.e., no evidence of eye rolls or blinks and
overall EMG output below 15 uV) was required for completion of
this assessment task. The next 90-s task was reading. Material that Table 2
was age or grade appropriate was selected (e.g., school reading Summary of all Analyses of Variance Effects
texts, and reading tasks from the Kaufman Test of Educational
Achievement (Kaufman & Kaufman, 1985), the Peabody Indi- Effect dfi, F P
vidual Achievement Test (Dunn & Markwardt, 1970), or other Diagnosis 2,580 29.47 <.001
age-related reading tests) and read silently by the participant. Age 3,580 28.95 <.001
Again, after completion of this task, the EEG was reviewed in 2-s Task 3, 1740 7.87 <.001
intervals to eliminate epochs with excessive EMG activity or eye Diagnosis X Age 6,580 2.96 <.001
movement or blink artifact. A minimum of 15 low-artifact epochs Diagnosis X Task 6, 1740 1.24 <.280
Age X Task 9, 1740 0.86 <.560
was required for completion of this assessment task. A 90-s Diagnosis X Age X Task 18, 1740 1.09 <.360
listening task occurred next. Age appropriate material was selected
428 MONASTRA ET AL.

Table 3 nonclinical control group. Although differentiation between

Mean Theta-Beta Power Ratios the two ADHD groups was suggested by the graphic
Diagnosis depiction of the mean theta-beta ratio data (see Figures
1-4), statistically significant differences were noted only on
Age (years) ADHD-I ADHD-H/C Control the drawing task. On the drawing task, mean power ratios for
6-11 8.485 7.698 3.027 the ADHD, combined group were significantly greater than
12-15 4.494 5.547 2.059 those demonstrated by the ADHD, inattentive group
16-20 3.617 4.188 1.999 (p < .01).
21-30 2.454 4.125 1.495
Note. ADHD-I = attention deficit hyperactivity disorder, inatten- The Theta-Beta Power Ratio as a Test for ADHD
tive; ADHD-H/C = attention deficit hyperactivity disorder, hyper-
active-combined type. The second hypothesis of this study was that critical
values derived from the means and standard deviations of
group (ages 6-11 years old). When compared with each of the theta-beta power ratio of the control groups could serve
the other age groups, the participants aged 6-11 years had as a basis for differentiating participants with ADHD from
significantly higher theta-beta ratios (Tukey HSD tests, nonclinical control participants. In order to define critical
p < .001) on each of the four tasks. Although continued values for ADHD, the mean theta-beta ratio was first
reduction in the theta-beta ratios was associated with calculated for each of the four control groups, collapsing
increased age, as reflected in Table 3 and Figures 1,2, 3, and across all tasks. Critical values for ADHD were defined as
4, post hoc comparison of these age effects was not 1.0, 1.5, and 2.0 SDs above the mean for each of the control
significant. The primary improvement in the level of cortical groups. A summary table of these critical values is provided
arousal was apparent by ages 12-15 years old and persisted in Table 4.
through the age of 30 years old. After calculating critical values for ADHD, an overall
Post hoc analysis of the effect of diagnostic classification power ratio score was derived for each participant. This ratio
revealed a consistent pattern of differentiation of both score was obtained by averaging the theta-beta power ratios
ADHD groups from the nonclinical control group on all for each participant on the four tasks. Participants were
tasks. Examination of between-group differences using classified as ADHD or non-ADHD on the basis of the power
Tukey's HSD test revealed statistically significant differen- ratio alone by using cutoffs of 1.0,1.5, and 2.0 SDs from the
tiation between both of the ADHD groups and the control mean of each of the nonclinical control groups. Because the
group on the baseline (p < .001), reading (p < .001), goal of this initial study was to examine whether an
listening (p < .001), and drawing (p < .001) tasks. Within- attentional index derived from QEEG data (i.e., the theta-
subject comparisons across task revealed that individuals beta power ratio averaged over four tasks) could differenti-
classified as either ADHD, inattentive or ADHD, combined ate individuals with ADHD from nonclinical controls,
type showed significantly higher power ratios on the draw- accurate classification was considered to occur when the
ing task relative to their ratios on the other tasks (p < .01). theta-beta ratio score was in agreement with classification as
No such pattern was observed in participants from the ADHD or non-ADHD on the basis of behavioral rating

Rao R (24,1588)=1.12; p<.3144

ill 7
<

DX
ADHD(I)
DX
ADHD(C)
DX
6-11 12-15 16-20 21-30 CONTROL

AGE

Figure 1. Plot of the mean theta-beta power ratios for the two-way interaction, Age (Years) X
Diagnosis, during the eyes-fixed baseline task. DX = diagnosis; ADHD(I) = attention deficit
hyperactivity disorder, inattentive type; ADHD(C) = attention deficit hyperactivity disorder,
hyperactive-combined type.
 ASSESSING ADHD VIA QEEG 429

Rao R (24,1588)=1.12; p<.3144

10
g
8
7

5
4

3
-o- DX
2 ADHD(I)
o DX
1 ADHD(C)
0 o DX
6-11 12-15 16-20 21-30 CONTROL

AGE

Figure 2. Plot of the mean theta-beta power ratios for the two-way interaction, Age (Years) X
Diagnosis, during the reading task. DX = diagnosis; ADHD(I) = attention deficit hyperactivity
disorder, inattentive type; ADHD(C) = attention deficit hyperactivity disorder, hyperactive-
combined type.

scales and continuous performance tests. False-positive groups was used as a critical value, the rate of diagnostic
classification occurred when the theta-beta ratio score agreement was above 85% for each group (M = 88%). At
indicated ADHD in a participant classified as non-ADHD on 1.5 SDs, the agreement rate ranged from 81% to 91%
the behavioral rating scales and continuous performance (M = 84%). At 2.0 SDs, the agreement rate dropped to 76%,
tests. False-negative classification occurred when the theta- with 23% of the errors resulting from false-negative ratings.
beta ratio score indicated non-ADHD in a participant Additional analysis of classification accuracy was con-
classified as ADHD in the screening process. A summary of ducted in order to examine the sensitivity and specificity of
the accuracy rates is provided in Table 5. the QEEG-derived attentional index. In this analysis, a
Examination of the accuracy rates provided in Table 5 participant whose attentional index was 1.5 SDs greater than
reveals a high degree of consistency between classification the mean of the age appropriate nonclinical control group
derived from our index of cortical slowing and those was considered positive for ADHD. Examination of the
obtained through behavioral rating scales and continuous percentage of participants classified with either type of
performance tests. When 1 SD above the mean for control ADHD who tested positive on the QEEG revealed a

Rao R (24,1588)=1.12; p<.3144

10
9

8
7

5
4

3
<> -o- DX
2 ADHD(I)
D DX
1 ADHD(C)
..„«_ DX
0 CONTROL
6-11 12-15 16-20 21-30
AGE

Figure 3. Plot of the mean theta-beta power ratios for the two-way interaction, Age (Years) X
Diagnosis, during the listening task. DX = diagnosis; ADHD(I) = attention deficit hyperactivity
disorder, inattentive type; ADHD(C) = attention deficit hyperactivity disorder, hyperactive-
combined type.
430 MONASTRA ET AL.

Rao R (24,1588)=1.12; p<.3144

-o- DX
ADHD(I)
D DX
ADHD(C)
--o- DX
6-11 12-15 16-20 21-30 CONTROL

AGE

Figure 4. Plot of the mean theta-beta power ratios for the two-way interaction, Age (Years) X
Diagnosis, during the drawing task. DX = diagnosis; ADHD(I) = attention deficit hyperactivity
disorder, inattentive type; ADHD(C) = attention deficit hyperactivity disorder, hyperactive-
combined type.

sensitivity rating of 86%. Specificity of the QEEG measure clinical researchers seeking to examine the validity of a
(i.e., the percentage of non-ADHD participants testing simplified QEEG indicator as a laboratory test for ADHD.
negative for ADHD) was 98%. The overall positive predic- The present study clarified certain electrophysiological
tive power of the measure was 99%, meaning that only 1% parameters and assessment procedures that can be used to
of the individuals who tested positive on the measure did not accurately classify ADHD patients and nonclinical controls.
have ADHD. Consequently, the results of our evaluation of The level of accuracy obtained using our neurometric
test sensitivity and specificity were considered supportive of indicator was similar to that presented by the developers of
the use of the theta-beta power ratio in assessing ADHD. behavioral and continuous performance tests for ADHD. In
addition, the present findings yielded levels of accuracy
Discussion similar to those reported by researchers using discriminant
function analysis of multichannel EEG recordings.
The essential findings of this study were as follows. First, These findings are consistent with the results of neurologi-
a significant association was noted between age and a cal assessment procedures (PET, SPECT, MRI), as well as
neurometric indicator of cortical slowing (the theta-beta emerging neuropsychologically based models associating
power ratio obtained at the vertex using a referential ADHD with prefrontal cortical functioning (Barkley, 1997b).
montage). Second, scores on this indicator were signifi- In addition, our findings, similar to those presented by Mann
cantly higher in patients with attention deficit disorders et al. (1992), Lubar (1995; Lubar et al., 1996), and Chabot
(both ADHD-I and ADHD-H/C) than nonclinical controls and his associates (Chabot et al., 1996; Chabot & Serfontein,
for ages 6 through 30 years old. Third, critical values derived 1996), are supportive of the development of QEEG-based
from the neurometric scores of the nonclinical controls assessment procedures for evaluating ADHD. Because the
could serve as a basis for accurate classification of the preponderance of neurological, biochemical, and electro-
participants of the study. Fourth, this indicator of cortical physiological research has supported the conclusion that
slowing yielded similar accuracy rates, regardless of the sex ADHD is a health impairment, it appears imperative that
of the participant. assessment procedures be developed to assess the physical
In summary, these findings provide initial guidelines for
Table 5
Table 4 Accuracy of Classification Using the
Critical Values for Attention Deficit Hyperactivity Disorder Theta-Beta Power Ratio
on the Basis of Power Ratios
Classification
SD Criterion Correct (%) False- (%)
False + (%)
Age (years) 1.0 1.5 2.0
1.0 SD 88 3 9
6-11 4.36 5.03 5.69 1.5SD 84 2 14
12-15 2.89 3.31 3.72 2.0 SD 76 1 23
16-20 2.24 2.36 2.48
21-30 1.92 2.13 2.34 Note. False+ = false-positive classification; False— = false-
negative classification.
 ASSESSING ADHD VIA QEEG 431

as well as the neuropsychological and behavioral symptoms poses. Finally, in order to assess test specificity, examination
of this disorder. of the accuracy of this indicator to differentiate conditions
Because QEEG procedures are relatively nonintrusive, such as oppositional defiant disorder or affective disorders
inexpensive, and can provide information about cortical from ADHD is required.
processes that are difficult to obtain from neuroimaging Comparisons with behavioral and cognitive tests (continu-
scans (e.g., degree of coherence and symmetry in activity ous performance tests) developed to assess ADHD likewise
between different cortical regions), their application in seem desirable in order to examine the issue of construct
developing an understanding of ADHD appears promising. validity. Although both types of measures were obtained in
QEEG researchers like Mann et al. (1992), Lubar (1995; our current study, they were used for classification purposes.
Lubar et al., 1996) and Chabot and his colleagues (Chabot et Consequently, comparisons between behavioral, cognitive,
al., 1996; Chabot & Serfontein, 1996) have shown that and neurometric measures were not made during this study
multichannel EEG recordings and an examination of QEEG because the degree of correlation would be artificially
characteristics, such as electrophysiological power, power inflated. However, such comparisons are planned in our
ratios, coherence, and symmetry, can be useful in differenti- ongoing research. Our goal remains not to supplant behav-
ating individuals with ADHD from nonclinical controls and ioral or cognitive measures but to add a neurometric
from peers with learning disorders. Our study sought to laboratory test to aid in the diagnostic process.
examine the sensitivity and specificity of a QEEG scan for Two additional research directions are derived from what
ADHD on the basis of the electrophysiological output from was not demonstrated in our study. Specifically, we are
a single channel recording at the vertex. aware that the current findings did not reveal significant
Similar to the findings of Mann et al. (1992), Lubar (1995; differences between the subtypes of ADHD on any task
Lubar et al., 1996), and Chabot and Serfontein (1996), the other than drawing and only examined individuals aged 6 to
results of our study provided further evidence of cortical 30 years. Several modifications in our approach to assess-
slowing in participants with ADHD. Mann et al. examined ment for subtypes and patients above the age of 30 years old
electrophysiological power from 19 sites and concluded that are planned.
participants with ADHD exhibited higher theta (4.00-7.75
In order to attempt differentiation of subtypes, analysis of
Hz) activity at several frontal and central locations. Lubar
the sensorimotor rhythm (12-15 Hz) is planned. Sensorimo-
(1995; Lubar et al., 1996) reported significantly higher
tor rhythm represents inhibitory activity generated in path-
theta-beta power ratios at several central and frontal loca-
ways originating in the cerebellum and terminating on motor
tions (including the vertex). Chabot and Serfontein reported
neurons in the sensorimotor cortex (Sterman, 1996). Be-
two neurophysiological subtypes for ADHD; one type was
cause sensorimotor rhythm training (Lubar, 1995) has
characterized by theta-alpha excess (with normal alpha
yielded positive results in the treatment of two primary
mean frequency), and the other type by theta-alpha excess
clinical features of ADHD-H/C (impulsivity and hyperactiv-
coupled with decreased alpha mean frequency. Again, the
ity), examination of this frequency band may prove useful in
primary locations of interest were within frontal and central
locations. Our finding of significantly higher theta-beta differentiating ADHD-H/C from ADHD-I subtypes. Like-
power ratios at the vertex and high rates of classification wise, because patients with ADHD, hyperactive or com-
accuracy using this neurometric is consistent with these bined type, show multiple indicators of impaired motor
findings and supports further examination of a simplified control (both in lack of motor inhibition and frequently in
scanning procedure for ADHD. impaired handwriting ability), replication of the current PSA
The current findings provide a first step in the identifica- procedure during performance of graphomotor tasks would
tion of a neurometric test for ADHD that is far less intrusive provide an indication of the consistency of present findings
and expensive than other procedures. Given our results, we across samples and contribute to an understanding of certain
hypothesize that the use of such an indicator, in conjunction of the neuropsychological differences between subtypes.
with behavioral and continuous performance test measures, Finally, in order to identify neurometric indicators for
will serve to increase overall diagnostic accuracy by reduc- ADHD in individuals beyond age 30 years old, improved
ing error rates associated with nonneurologically based methods for initial classification seem required. Specifically,
conditions with similar behavioral symptoms. In order to behavioral assessment procedures for individuals over the
continue the process of test development, a series of studies age of 18 years old have typically relied exclusively on
is required. self-report. As indicated by Barkley (1997a), the self-report
First, to ensure valid comparisons across clinical research of individuals with ADHD may underestimate symptom
centers, standardization of the assessment process is re- severity. Consequently, the inclusion of ratings provided by
quired. This will necessitate the development of software relatives (e.g., using the ADDES, Adult Version; McCarney,
programs for stimuli presentation, as well as neurometric 1996) and an examination of a large sample size of adults
data processing. Second, issues of test-retest reliability need may prove useful in determining whether our neurometric
to be addressed. Third, examination of the ability of the index of cortical slowing will continue to differentiate
neurometric assessment process developed in this study to persons with ADHD through adulthood. In addition, expan-
accurately classify participants not involved in our initial sion of our neurometric examination to include other
standardization study is needed for cross-validation pur- slow-wave frequencies (e.g., 6-10 Hz) is planned.
432 MONASTRA ET AL.

References children with attention deficit or specific developmental learning

disorders. Clinical Electroencephalography, 27, 26-34.
Achenbach, R. M., & Edelbrock, C. S. (1983). Manual for the Chabot, R. A., & Serfontein, G. (1996). Quantitative electroencepha-
Child Behavior Checklist and Revised Child Behavior Profile. lographic profiles of children with attention deficit disorder.
Burlington: University of Vermont, Department of Psychiatry. Biological Psychiatry, 40, 951-963.
Amen, D. G., Paldi, J. H., & Thisted, R. A. (1993). Evaluating Conners, C. K. (1973). Rating scales for use in drug studies with
ADHD with brain SPECT imaging. Journal of the American children. Psychopharmacology Bulletin, 9, 24-84.
Academy of Child and Adolescent Psychiatry, 32 (5), 1080- Conners, C. K. (1994). Conners' Continuous Performance Test.
1081. Toronto, Canada: Multi-Health Systems.
American Psychiatric Association. (1980). Diagnostic and statisti- Cooley, J. W., & Tukey, J. W. (1965). An algorithm for the machine
cal manual of mental disorders (3rd ed.). Washington, DC: calculation of complex Fourier series. Mathematics of Computa-
Author. tion, 19, 267-301.
American Psychiatric Association. (1987). Diagnostic and statisti- Dunn, L. M., & Markwardt, F. C. (1970). The Peabody Individual
cal manual of mental disorders (3rd ed., rev.). Washington, DC: Achievement Test. Circle Pines, MN: American Guidance Ser-
Author. vice.
American Psychiatric Association. (1994). Diagnostic and statisti- DuPaul, G. J. (1991). ADHD Rating Scale. In R. A. Barkley (Ed.),
cal manual of mental disorders (4th ed.). Washington, DC: Attention-deficit hyperactivity disorder: A clinical workbook (pp.
Author. 46^8). New York: Guilford Press.
Andreassi, J. (1989). Psychophysiology, human behavior and Gordon, M. (1983). The Gordon Diagnostic System. Dewitt, NY:
physiological response. New York: Erlbaum. Gordon Systems.
Barkley, R. A. (1990). Attention-deficit hyperactivity disorder: A Greenberg, L. M. (1994). Test of Variables of Attention Continuous
handbook for diagnosis and treatment. New York: Guilford Performance Test. Los Alamitos, CA: Universal Attention
Press. Disorders.
Barkley, R. A. (199la). ADHD Clinical Interview (Parent or Adult Hynd, G. W, Hern, K. L., Novey, E. S., Eliopulos, D., Marshall, R.,
Form). In R. A. Barkley (Ed.), Attention-deficit hyperactivity
Gonzalez, J. J., & Voeller, K. K. (1993). Attention deficit-
disorder: A clinical workbook (pp. 5-43). New York: Guilford
hyperactivity disorder and asymmetry of the caudate nucleus.
Press.
Journal of Child Neurology, 8, 339-347.
Barkley, R. A. (1991b). The ecological validity of laboratory and
Hynd, G. W., Semrud-Clikeman, M., Lorys, A. R., Novey, E. S.,
analogue assessment methods of ADHD symptoms. Journal of
Eliopulos, D., & Lyytinen, H. (1991). Corpus callosum morphol-
Abnormal Child Psychology, 19, 149-178.
ogy in attention deficit-hyperactivity disorder: Morphometric
Barkley, R. A. (1997a). Age dependent decline in ADHD: True
analysis of MRI. Special Series: Attention deficit disorder.
recovery or statistical illusion? The ADHD Report, 5 (1), 1-5.
Barkley, R. A. (1997b). Behavioral inhibition, sustained attention, Journal of Learning Disabilities, 24, 141-146.
and executive functions: Constructing a unifying theory of Janzen, T, Graap, K., Stephanson, S., Marshall, W., & Fitzsim-
ADHD. Psychological Bulletin, 121, 65-94. mons, G. (1995). Differences in baseline EEG measures for
Barkley, R. A., Grodzinsky, G., & DuPaul, G. J. (1992). Frontal ADD and normally achieving preadolescent males. Biofeedback
lobe functions in attention deficit disorder with and without & Self Regulation, 20, 65-82.
hyperactivity: A review and research report. Journal of Abnor- Kaufman, A. S., & Kaufman, N. L. (1985). The Kaufman Test of
mal Child Psychology, 20, 163-188. Educational Achievement. Circle Pines, MN: American Guid-
Beery, K. E., & Buktenica, N. A. (1967). The Beery Developmental ance Service.
Test of Visual-Motor Integration. Columbus, OH: Modern Cur- Klinkerfuss, G. H., Lange, P. H., Weinberg, W. A., & O'Leary, J. L.
riculum Press. (1965). Electroencephalographic abnormalities of children with
Benton, A. L. (1955). Visual Retention Test. New York: The hyperkinetic behavior. Neurology, 15, 883-891.
Psychological Corporation. Kuperman, S., Johnson, B., Arndt, S., Lindgren, S., & Wolraich, M.
Biederman, J., Faroane, S. V., Spencer, R., Wilens, R., Norman, D., (1996). Quantitative EEG differences in a nonclinical sample of
Lapey, K. A., Mick, E., Krifcher-Lehman, B., & Doyle, A. children with ADHD and undifferentiated ADD. Journal of the
(1990). Patterns of psychiatric comorbidity, cognition and psy- American Academy of Child and Adolescent Psychiatry, 35,
chosocial functioning in adults with attention deficit disorder, 1009-1016.
residual type. Comprehensive Psychiatry, 31, 416^4-25. Levy, F. (1994). Neurometrics: Review and comments. The ADHD
Cantwell, D. P. (1996). Attention deficit disorder: A review of the Report, 2(5), 1-3.
past ten years. Journal of the American Academy of Child and Linden, M., Gevirtz, R., Isenhart, R., & Fisher, T. (1996). Event
Adolescent Psychiatry, 35, 978-987. related potentials of subgroups of children with attention deficit
Capute, A. J., Niedermeyer, E. F. L., & Richardson, F. (1968). The hyperactivity disorder and the implications for EEG biofeed-
electroencephalogram in children with minimal brain dysfunc- back. Journal of Neurotherapy, 1, 1-11.
tion. Pediatrics, 4, 1104-1114. Loiselle, D. I., Stamm, J. S., Maitinisky, S., & Whipple, S. C.
Casey, B. J., Castellanos, F. X., Giedd, J. N., Marsh, W. L., (1980). Evoked potential and behavioral signs of attentive
Hamburger, S. D., Schubert, A. B., Vauss, Y. C., Vaituzis, A. C., dysfunctions in hyperactive boys. Psychophysiology, 17, 193-
Dickstein, D. P., Sarfatti, S. E., & Rapoport, J. L. (1997). 201.
Implication of right frontostriatial circuitry in response inhibi- Lou, H. C., Henriksen, L., & Bruhn, P. (1984). Focal cerebral
tion and attention-deficit/hyperactivity disorder. Journal of the hypoperfusion in children with dysphasia and/or attention deficit
American Academy of Child and Adolescent Psychiatry, 36(3), disorder. Archives of Neurology, 41, 825-829.
374-383. Lubar, J. F. (1991). Discourse on the development of EEG
Chabot, R. A., Merkin, H., Wood, L. M., Davenport, T. L., & diagnostics and biofeedback for attention-deficit/hyperactivity
Serfontein, G. (1996). Sensitivity and specificity of QEEG in disorders. Biofeedback & Self Regulation, 16, 201-225.
 ASSESSING ADHD VIA QEEG 433

Lubar, J. F. (1995). Neurofeedback for the management of attention Rie, H. E., & Rie, E. D. (Eds.). (1980). Handbook of minimal brain
deficit hyperactivity disorders. In M. S. Schwartz (Ed.), Biofeed- dysfunction: A critical view. New York: Wiley.
back: A practitioners guide (pp. 493-522). New York: Guilford Sanford, J. A. (1994). Intermediate Visual and Auditory Continuous
Press. Performance Test. Richmond, VA: BrainTrain.
Lubar, J. R, Swartwood, M. O., Swartwood, J. N., & Timmermann, Satterfield, J. H., Schell, A. M., Nicholas, T. W., Satterfield, B. T, &
D. L. (1996). Quantitative EEG and auditory event-related Freese, T. E. (1990). Ontogeny of selective attention effects on
potentials in the evaluation of attention-deficit/hyperactivity event-related potentials in attention-deficit hyperactivity disor-
disorder: Effects of methylphenidate and implications for neuro- der and normal boys. Biological Psychiatry, 28, 879-903.
feedback training. Journal of Psychoeducational Assessment, StatSoft. (1995). Statistica. Tulsa, Oklahoma: StatSoft.
143-160. Sterman, M. B. (1996). Physiological origins and functional
Mann, C, Lubar, J., Zimmerman, A., Miller, C, & Muenchen, R. correlates of EEG rhythmic activities: Implications for self-
(1992). Quantitative analysis of EEG in boys with attention- regulation. Biofeedback and Self-Regulation, 21, 3-33.
deficit-hyperactivity disorder: Controlled study with clinical Trommer, B. L., Hoeppner, J. B., Lorber, R., & Armstrong, K.
implications. Pediatric Neurology, 8, 30-36. (1988). Pitfalls in the use of a continuous performance test as a
Mannuzza, S., Klein, R. G., Bessler, A., Malloy, P., & LaPadula, M. diagnostic tool in attention deficit disorder. Journal of Develop-
(1993). Adult outcome of hyperactive boys: Educational achieve- mental and Behavioral Pediatrics, 9, 339-345.
ment, occupational rank and psychiatric status. Archives of Ulmann, R. K., Sleator, E. K., & Sprague, R. (1984). Anew rating
General Psychiatry, 50, 565-576. scale for diagnosis and monitoring of ADD children. Psychophar-
Mannuzza, S., Klein, R. G., Bonagura, N., Malloy, P., Giampino, macology Bulletin, 20, 160-164.
H., & Addalli, K. A. (1991). Hyperactive boys almost grown up: Zametkin, A. J., Nordahl, T. E., Gross, M., King, A. C., Semple,
Replication of psychiatric status. Archives of General Psychia- W. E., Rumsey, J., Hamburger, S., & Cohen, R. M. (1990).
try, 48, 77-83. Cerebral glucose metabolism in adults with hyperactivity of
McCarney, S. B. (1989). Attention Deficit Disorders Evaluation childhood onset. New England Journal of Medicine, 323,
Scale. Columbia, MO: Hawthorne Press. 1361-1366.
McCarney, S. B. (1996). The Adult Attention Deficit Disorders Zametkin, A. J., & Rapoport, J. L. (1987). Noradrenergic hypoth-
Evaluation Scale. Columbia, MO: Hawthorne Press. esis of attention deficit disorder with hyperactivity: A critical
McCarthy, D. (1972). McCarthy Scales of Children's Abilities. review. In H. V. Metsler (Ed.), Psychopharmacology: The third
New York: The Psychological Corporation. generation of progress (pp. 837-842). New York: Raven Press.
Morrison, J. R. (1980). Childhood hyperactivity in an adult
psychiatric population: Social factors. Journal of Clinical Psy-
chiatry, 41, 40-43.
Rezai, K., Andreasen, N. C., Alliger, R., Cohen, G., Swayze, V., & Received September 5, 1997
O'Leary, D. S. (1993). The neuropsychology of the prefrontal Revision received September 18, 1998
cortex. Archives of Neurology, 50, 636-642. Accepted November 23, 1998

Low Publication Prices for APA Members and Affiliates

Keeping you up-to-date. All APA Fellows, Members, Associates, and Student Affiliates
receive—as part of their annual dues—subscriptions to the American Psychologist and
APA Monitor. High School Teacher and International Affiliates receive subscriptions to
the APA Monitor, and they may subscribe to the American Psychologist at a significantly
reduced rate. In addition, all Members and Student Affiliates are eligible for savings of up
to 60% (plus a journal credit) on all other APA journals, as well as significant discounts on
subscriptions from cooperating societies and publishers (e.g., the American Association for
Counseling and Development, Academic Press, and Human Sciences Press).

Essential resources. APA members and affiliates receive special rates for purchases of
APA books, including the Publication Manual of the American Psychological Association,
and on dozens of new topical books each year.

Other benefits of membership. Membership in APA also provides eligibility for

competitive insurance plans, continuing education programs, reduced APA convention fees,
and specialty divisions.

More information. Write to American Psychological Association, Membership Services,

750 First Street, ME, Washington, DC 20002-4242.