Quantitative electroencephalographic profiles of children with attention deficit disorder

Quantitative Electroencephalographic Profiles of Children with Attention Deficit Disorder Robert J. Chabot and Gordon Serfontein

Quantitative electroencephalogram (QEEG) was obtained from 407 children with attention deficit disorder. These QEEGs were compared to those of 310 normal children. Discriminant analysis resulted in a specificity of 88% and a sensitivity of 93. 7% for distinguishing normal children from those with attention deficit disorder. Two major neurophysiological sub~pes were evident within the 92.6% abnormal QEEG profiles encountered. The first showed varying degrees of EEG slowing, especially in frontal regions, whereas the second showed an increase in EEG activity, especially in frontal regions. Deviations from normal development rather than maturational lag were present as the source of the neurophysiological abnormality in the majority of these children. In conjunction with recent magnetic resonance imaging, positron emission tomography, and regional cerebral blood flow studies, these results indicate neurophysiological dysfunction within the cortical and subcortical structures that serve the frontal/striatal system. Models suggesting both hypo- or hyperarousal of these structures are supported. © 1996 Society of Biological Psychiatry Key Words:

Quantitative EEG, neurometrics, attention deficit disorder

BIOL PSYCHIATRY 1996;40:951--963

Introduction Differential diagnosis and treatment of children presenting with possible attention deficit disorder with (ADHD), or without hyperactivity (ADD), and/or learning problems can be difficult, especially when only clinical information about behavior and learning is available. Behavioral symptoms of hyperactivity, inattention, and/or impulsivity are the most common features of this class of disorders, which has an estimated prevalence of 6% within our society (Schachar 1991). The broad nature of current diagnostic From the Brain Research Laboratory, Department of Psychiatry, New York University Medical Center, New York, New York (RJC); and The Serfontein Clinic of Developmental Paediatrics and Learning Disorders, Sydney, Australia (RJC, GS). Address reprint requests to Robert J. Chabot, PhD, Brain Research Labs, 8th Floor, Old Bellevue Administration Bldg.. 27th St. at 1st Ave., New York, NY 10016. Received March 6, 1995; revised October 16, 1995.

© 1996 Society of Biological Psychiatry

criteria used to assess ADD/ADHD suggests that a heterogeneous population of children may be subsumed within this disorder (Mann et al 1992; Weinberg and Brumback 1992). Small sample sizes and restricted patient sampling procedures make it difficult to generalize research findings about possible central nervous system (CNS) dysfunction underlying this class of behavioral disorders. Studies examining a large cohort of children showing the entire range of attention, hyperactivity, and impulsivity symptoms are needed to identify a possible pathophysiological substrate that might serve to define this population of children. Previous electroencephalographic (EEG) research has been inconclusive in documenting both the prevalence and the nature of any underlying neurophysiological dysfunction in ADD/ADHD children. Studies using conventional EEG report abnormal findings in between 30 and 60% of 0006-3223/96/$15.00 SSDI 0006-3223(95)00576-5

952

BIOLPSYCHIATRY

R.J. Chabot and G. Serfontein

1996;40:951-963

children with ADHD (Small 1993), although a recent paper found abnormal EEG in only 1 of 11 children with ADHD (Phillips et al 1993). Other studies examined selected quantitative EEG features in small samples of hyperactive boys within the age range of 9-12 years. Findings of decreased alpha and/or beta power, increased left posterior theta and delta power, and increased withinhemisphere coherence have been reported (Callaway et al 1983; Dykman et al 1982; Montagu 1975; Matsuura et al 1993. However, these studies used EEG recording montages restricted to central, parietal, and/or occipital regions. In the two studies that used more extensive recording montages, topographic differences in EEG abnormality were observed. An increased frontal theta/beta power ratio was reported in boys with attention problems without hyperactivity (Lubar 1991). Mann and associates found increased theta (greater in frontal regions) and decreased betal (especially in temporal regions) in a sample of 25 ADD boys (Mann et al 1992). In the present prospective study we use neurometric quantitative EEG (QEEG) to document the prevalence and nature of neurophysiological dysfunction in a large population of male and female children with attention and learning problems that span the behavioral space defined by symptoms of inattention, hyperactivity, and impulsivity. This technique quantifies EEG recorded across 19 regions of the International 10/20 system, and has been shown to be a sensitive indicator of cortical electrophysiological dysfunction in children and adults with neurological and psychiatric disorders (John et al 1983, 1992; Prichep and John 1992). It has also been shown to be more sensitive than conventional EEG, magnetic resonance imaging (MRI), and measures of cerebral blood flow for identifying functional CNS disturbance (John et al 1988; Jonkman et al 1985; Ritchlin et al 1992). QEEG should prove useful in determining whether any common underlying source of brain dysfunction is present in children with ADD/ ADHD, and whether children with various combinations of the behavioral symptoms associated with this class of disorders can be distinguished from one another. Thus, in the present study we include ADD children with and without hyperactivity, with and without learning problems, as well as children referred for attention problems that fail to reach DSM-III criteria for ADD.

Methods

sample of 439 children seen between June 1991 and December 1992 were entered into this study. We were able to obtain reliable EEG recordings on 407 of these children. All children were examined by a pediatric neurologist, and had a neuropsychological and QEEG evaluation. None of the children was receiving medication at the time of testing. Children with histories of epilepsy, drug abuse, head injury, or psychotic disorders were excluded. The clinical and neuropsychological evaluations obtained on each child included the following: 1) Connors ParentTeacher Questionnaire, completed by the child's primary teacher for those children in school and by a parent otherwise; 2) DSM-III Symptom List rating scale with ratings of hyperactivity, inattention, impulsivity, and peer interactions completed by the primary teacher, if in school, and by a parent otherwise; 3) a 4-point scale rating memory problems involving reading, spelling, math, and other areas; 4) the Wechsler Intelligence Scale for Children-Revised (WISC-R) (children under 17) or Wechsler Adult Intelligence Scale--Revised (WAIS-R) (children 17 and over) intelligence battery; and 5) the Neale Analysis of Reading Ability, yielding scores of reading rate, accuracy, and comprehension. In 70 children the WISC-R battery could not be administered and an estimate of full-scale IQ was obtained using the Slosson Intelligence Test. In younger children with minimal reading skills, the Schonell Graded Word Reading Test was administered instead of the Neale test.

Normal Population The normal population included 310 children between the ages of 6 and 17 years. All "normal" subjects were free of neurological or medical disease, had no history of head injury or drug or alcohol abuse, were of normal IQ, showed evidence of adequate functioning at home/school for the past 2 years, and had not taken any prescription medication for at least 90 days prior to evaluation. Specific details of the procedures used to construct normal QEEG age-regression equations from this database have been previously published (John et al 1980). The reliability of this normal database has been validated both within individual patients (John et al 1983; Kaye et al 1981), and across independent samples of normal individuals (Alvarez et al 1987, 1989; Gasser et al 1983; Harmony 1988; Jonkman et al 1985; Ritchlin et al 1992; Veldhuizen et al 1993; Yingling et al 1986). The independent replications of the age-regression equations developed on the above database justify their generalized application (Lopes da Silva 1990).

Clinical Population

Quantitative EEG (QEEG) Methodology

All children were referred to the Developmental Paediatrics and Learning Disorders Clinic in Sydney, Australia. A

The neurometric method of QEEG data collection and analysis was utilized (John et al 1988). Patients were

QEEG in Children with Attention Problems

BIOLPSYCHIATRY

953

1996;40:951-963

seated comfortably in a sound- and light-attenuated room during the evaluation. Electrode caps were used to place recording electrodes over the 19 standard regions defined by the International 10/20 system referenced to linked ears. All electrode impedance levels were kept below 5000 12. The EEG amplifiers had a ban@ass from 0.5 to 70 Hz (3 dB points), with a 50-Hz notch filter (all data collected in Australia). Eight additional bipolar channels were derived during analysis. These included the left and right frontal/temporal, central, temporal, and parietal/occipital regions. Eye movement and the electrocardiogram were also recorded to aid in selecting artifact-free segments of EEG for subsequent quantitative analysis. Twenty to thirty minutes of continuous eyes closed resting EEG was recorded using a Spectrum 32 QEEG system (Cadwell Laboratories, Kennewick, WA). An experienced EEG technician observed the continuous EEG being recorded and selected from 24 to 48 artifact-free epochs of EEG, each of 2.5 sec duration. The exact number chosen depended upon how cooperative the child was, with a minimum of 24 epochs required for study entry. These EEG epochs were digitized and placed onto floppy disks for entry into a SUN work station for further analysis. Prior to EEG quantification, all epochs were reviewed by a second, independent experienced EEG technician, who removed artifact-contaminated epochs missed by the first technician. Subsequent analyses were restricted to those children from whom a minimum of 1 rain of artifact-free EEG could be obtained. This is the minimum amount of EEG required to obtain reliable quantitative measures (John et al 1987). Reliable EEG recordings were obtained from 407 children with an average of 71.8 sec of EEG from each child (range = 60-117.5 sec). The artifact-free EEG from each channel was converted from the time to the frequency domain via fast Fourier transform (FFT). The results of the FFT were used to calculate the following QEEG measures: (1) absolute power, the amount of energy within the delta, theta, alpha, and beta frequency bands, as well as total power (summed across frequency bands) for each monopolar and bipolar channel; (2) relative power, the percentage of total power within each frequency band for each monopolar and bipolar channel; (3) interhemisphere power asymmetry, a ratio of the absolute power within each frequency band and for total power calculated between eight monopolar (Fpl/Fp2, F3/F4, F7/F8, C3/C4, T3/T4, T5/T6, P3/P4, O1/O2) and four bipolar regions (left and right frontal/temporal, temporal, central, and parietal/occipital); (4) interhemisphere waveshape coherence, the cross-correlation of EEG waveforms in each frequency band independent of power, calculated between the eight monopolar and four bipo-

lar regions described above; (5) intrahemispheric power asymmetry, a ratio of absolute power within each frequency band and for total power calculated between frontal/temporal (F3/T5, F7/T5, F4/T6, and F8/T6) and frontal/occipital regions (F3/O1, F7/O1, F4/O2, and F8/O2) within each hemisphere; (6) intrahemispheric waveshape coherence, the cross-correlation of EEG waveforms in each frequency band independent of power, calculated between the eight monopolar intrahemispheric regions described above; and (7) mean frequency, the frequency of the EEG above and below which half the power lies, calculated within each frequency band and for total power for all monopolar and bipolar recordings. Two additional features were calculated for each recording site, maturational lag and developmental deviation (John et al 1983). A significant (Z > 1.96) score for a maturational lag measure indicates an abnormal finding that would be normal in a younger child. A significant score for a developmental deviation measure indicates an abnormal finding that would not be normal at any age.

QEEG Scoring Criteria Each QEEG measure was compared to the mean and standard deviation of that measure obtained from the age-regressed normal database using a Z or standard score [Z = (patient value - normal mean)/(normal standard deviation)]. This Z-score value is proportional to the probability of obtaining the patient value from the normal population. For example, there is a 5% chance that a Z score of + 1.96 will be obtained from a normal individual. The presence or absence of a QEEG abnormality, the nature of the QEEG abnormality, and whether the abnormality was focal or generalized was determined. A QEEG was within normal limits if less than 5% of the QEEG measures calculated reached the .05 level of statistical significance, and if there was no localized pattern to any abnormal Z scores present. A QEEG was called abnormal if a significant frequency and/or inter- or intrahemispheric abnormality was present. Similar criteria for scoring QEEG have previously been documented (John etal 1989; Oken e t a l 1989; Ritchlin et al 1992).

Results

Clinical Features of Patient Population To minimize the influence of IQ as a confounding variable, we placed the children into either a normal-IQ group (IQ > or = 85; n = 319), or a low-IQ group (IQ < 85; n = 88). Within each IQ group, the DSM-III rating scales of inattention, hyperactivity, and impulsive behavior were used to classify the children into the following groups: 1)

954

BIOLPSYCHIATRY

R.J. Chabot and G. Serfontein

1996;40:951-963

Table 1. Clinical Differences between Subgroups of Attention Deficit Disorder Children with Normal IQ Scores Learning disorder present (n = 149)

No learning disorder (n = 170)

Clinical variables % boys Mean age

ADDHY

ADD

ATT

Main effect

ADDHY

ADD

ATT

Main effect

(n = 82)

(n = 62)

(n = 26)

of L D

(n = 58)

(n = 66)

(n = 25)

of A D D

88.7 9.9

87.1 10.9

Range age Grade

6-16.7 4.3

6-16.3 5.3

Inattention

10.3

8.5

3.0

Impulsivity

11. l

7.7

1.4

8.8

2.2

0.64

8.6 18.5

4.2 11.0

2.0 3.9

Hyperactivity Peer interaction Connors Memory

73.1 9.9 6.6-14.9 4.2

***

94.9 12.1

80.3 11.6

72.0 12.0

***

6-16.8 6.3

6-14.4 5.8

7-16.5 6.2

11.4

8.4

3.7

***

11.4

7.2

1.3

***

9.2

2.0

0.36

***

8.0 19.9

3.7 11.4

1.3 4.1

*** *** ***

*

4.6

5.0

2.6

***

6.1

6.4

4.6

Verbal IQ Performance IQ

100.9 99.9

101.7 101.4

103.5 102.9

***

95.4 98.5

92.4 98.2

96.7 102.1

IQ Read rate b Read accuracyb

102.7 -0.25 -0.87

103.3 -0.26 -1.10

104.5 -0.76 -1.00

*** *** ***

97.1 -2.2 -3.1

95.7 -2.2 3.1

101.2 -2.8 3.3

Read comprehension b

-0.48

-0.67

-0.58

***

-3.1

-3.0

-3.0

*

"Separate Verbal IQ and Performance IQ scores were not available on all children. IQ score means based upon greater N. bChronological age - reading age. ANOVA significance: * = .01; ** - .001; *** = .0001. ADDHY = attention deficit disorder with hyperactivity; ADD - attention deficit disorder without hyperactivity; ATT = attention problems but does not reach criteria for ADD; LD - learning disorder.

those reaching DSM-III criteria for attention deficit disorder with hyperactivity (ADHD); 2) those reaching DSMIII criteria for attention deficit disorder without hyperactivity (ADD); and 3) those children not reaching criteria for ADD or ADHD. These children were characterized by inattentive behavior without signs of impulsive or hyperactive behavior, and are referred to as the attention problem group (ATT). Finally, each child was classified in terms of whether or not a learning disorder in math or reading was present (LD and no-LD). A learning disorder was present if the child's reading comprehension score was 2 or more years below chronological age, and/or if the arithmetic score on the IQ test was greater than 1.5 standard scores below full-scale IQ (August and Garfinkel 1989; Semrud-Clikeman et al 1992). Table 1 presents group means and the significance of group differences for each clinical measure for the children with normal IQ scores. Across all groups, the percentage of boys was greater than that of girls, with this difference smallest within the group with only attention problems. The children with learning disorders were 1.6 years older and 1.4 school grade levels higher than the children without learning problems. There was clear separation between the groups on the Connors and DSM-III rating scales of inattention, impulsivity, and hyperactivity. On these measures, the children in the ADHD subgroup scored higher than the ADD children, who scored higher than the ATT children. The children without a learning disorder scored slightly higher on the hyperactivity scale

than did the LD children. The LD children scored lower than the no-LD children on the memory scale, Verbal, and full-scale IQ, but not Performance IQ. Table 2 presents the clinical data for the children with

Table 2. Clinical Differences between subgroups of Attention Deficit Disorder Children with Low IQ Scores ADDHY

ADD

ATT

Main effect

(n = 39)

(n = 37)

(n = 12)

of A D D

% boys Mean age Range age Grade Inattention Impulsivity Hyperactivity Peer interaction Connors Memory Verbal IQ

79.5 11.5 6-16.7 5.5 11.4 11.4 8.6 8.5 18.6 7.5 79.0

81.1 12.1 6.7-16.9 6.2 8.6 7.1 1.9 3.9 9.4 7.1 77.0

50.0 12.2 7.3-16.1 6.1 3.9 1.2 1.5 1.5 4. I 5.8 77.8

Performance IQ IQ" Read rate h Read accuracy b

79.4 76.6 2.7 -3.8

77.5 75.3 3.4 -4.1

79.1 76.4 -2.6 3.0

Read comprehension ~'

-3.6

-4.1

-3.9

Clinical variables

"Separate Verbal IQ and Performance IQ scores were not available on all children. IQ score means based upon greater N. hChronological age - reading age. ANOVA significance: *** - .0001. ADDHY = attention deficit disorder with hyperactivity; ADD = attention deficit disorder without hyperactivity; ATT = attention problems but does not reach criteria for ADD.

QEEG in Children with Attention Problems

BIOLPSYCHIATRY

955

1996:40:951 963

IQ scores below 85. All children in this group had a learning disorder in math and/or reading. None of the differences reached significance for gender, age, or grade level. The ADHD group scored higher than the ADD group, which scored higher than the ATT group on the Conners and on the inattention, impulsivity, and peer interaction measures. The children in the ADHD group scored higher on the hyperactivity index than did the other two groups.

QEEG Abnormalio, in Children with Attention Deficit Disorder Figure 1 presents group average color-coded topographic maps of monopolar absolute power, relative power, mean frequency, interhemispheric power asymmetry and coherence, developmental deviation, and maturational lag. These topographic maps were averaged across all 407 children because the patterns of QEEG abnormality found were similar for all clinical groups, with differences reflecting mainly the degree rather than the type of abnormality present. The QEEG profile of children with attention problems is characterized by increased theta absolute and relative power, especially in frontal derivations, slight elevations in alpha relative power, and diffuse decreases in alpha and beta mean frequency. Interhemispheric abnormality includes parietal and posterior temporal asymmetry, marked frontal and posterior hypercoherence, and moderate central and parietal incoherence. Intrahemispheric findings include frontal/temporal and frontal/occipital asymmetry, frontal/temporal hypercoherence, and frontal/occipital incoherence. These patterns of QEEG abnormality represent deviations from normal development rather than maturational lag or a delay in normal development.

QEEG Discriminant Findings Stepwise discriminant analysis was used to compare the normal population of children with the children with attention problems who had normal IQ scores. This discriminant function was developed using a random split-half of both groups of children. For the children with attention problems each half included equal numbers of ADHD, ADD, and ATT children. This discriminant function was tested against the children in the independent split-halves of each group and the children with attention problems and IQ scores below 85. To meet the 10:1 subject to variable ratio assumption for replicative, conservative statistical discriminant analyses, QEEG variables entered into the discriminant function were preselected based upon t-test comparisons and the intercorrelations of the QEEG variables selected for possible entry (Weiner and Dunn 1966).

A discriminant function utilizing nine QEEG variables resulted in 94.8% correct classification of normal children and 93.1% correct classification of the normal IQ children with attention problems. Split-half replication resulted in a specificity of 88% (normal children called normal) and a sensitivity of 93.7% (ADHD, ADD, ATT children called abnormal). A total of 95.4% of the low-IQ group of children with attention problems was classified as abnormal. Variables entering into this discriminant function reflect the QEEG results depicted in Figure 1.

Frequency of QEEG Abnormal Findings Table 3 presents the frequency of occurrence of the types of QEEG abnormality present within the ADD, ADHD, and ATT normal- and low-IQ children. Abnormal QEEG profiles were found in 92.5% of the normal-IQ and 95.5% of the low-IQ children. The most common frequency abnormality was a theta excess present in 37.3% of the normal-IQ and in 40.9% of the low-IQ children. Theta excess of a generalized nature was in most cases greatest over frontal regions, with focal theta excess localized within frontal and/or midline regions 92% of the time. An alpha excess was the next most common abnormality. Focal alpha excess was localized within posterior and/or midline regions 84.1% of the time. Beta excess occurred in 13.1%, with focal beta excess localized in frontal and/or posterior regions 73.9% of the time. Delta excess was quite rare and when present was localized to posterior regions 70% of the time. Table 4 presents the frequency of occurrence of significant power asymmetry both within and between cortical hemispheres. Interhemispheric power asymmetry was present in 30.4% of the normal-IQ and 42.1% of the low-IQ children. A negative asymmetry was 6.2 times more likely to be present than a positive asymmetry, which indicates a greater right/left hemisphere power ratio than is seen in the normal population. Low-IQ children showed more negative asymmetry than did the normal-IQ children, with no differences in the frequency of occurrence of positive asymmetry. Significant intrahemispheric power asymmetry occurred more fi'equently in the low-IQ children than in the normal-IQ children (65.9% vs. 47.3%). Positive asymmetry (increased frontal power relative to temporal or occipital power) was 7.8 times more likely than negative asymmetry (increased temporal or occipital power relative to frontal power), with both types of abnormality equally likely to occur within the left or right hemispheres. Table 5 presents the frequency of occurrence of significant coherence abnormality both within and between cortical hemispheres. Significant incoherence represents decreased synchronization of the EEG across cortical

956

BIOL PSYCHIATRY

R.J. Chabot and G. Serfontein

1996;40:951-963

Gl'oup Aw:;rage Topogral)hic Maps of Monopolar QEEG Measures Averaged Across All Clnldrc. Tot, al

Delta

I h ~t;l

Alph.

13-t~1

I~OWEtl

Ml-;an I ~ [~ I";q I; I,;N( Y

Pow'H~ A~YMMICIItY

Q I!: I,X;

l{ I'll,A1 IV F I~oWt,:H

] }1' ~ I' I, ~t'Mf' ~, I A I

Figure 1. Group average topographic maps of Z-transformed monopolar absolute power, mean frequency, power asymmetry, and coherence for the total frequency distribution and separately for the delta, theta, alpha, and beta frequency bands (rows 1-4). Also present are topographic maps of Z-transformed relative power for each frequency band (row 5), and maps of developmental deviation and maturational lag calculated across all frequency bands. All topographic maps were averaged across all 407 children within our ADHD population with the color-coded values resulting from interpolation across the 19 monopolar recording sites. The top of each map represents the front of the head. All maps represent the mean Z-score deviation of the ADHD children from the normal population of children. Color-coding is proportional to the mean Z score of our ADHD sample with the significance of mean Z scores determined by taking into account the square root of the sample size. Thus, the 1% level of significance for a group of 400 patients would be equal to approximately (2.57/20) = .13. The scale for each topographic map is + 1.5 to - 1 . 5 Z-score units.

QEEG in Children with Attention Problems

BIOLPSYCHIATRY

957

1996;40:951-963

Table 3. Frequency of QEEG Profile Types in Attention Deficit Disorder Children Normal

Delta excess

Theta excess

A l p h a excess

Beta excess

Slow A M F

Clinical groups

Freq

All

Global

Focal

Global

Focal

Global

Focal

Global

Focal

Global

Focal

Interhem only

N o r m a l IQ (n = 319) L o w IQ (n = 88) Overall (n = 407)

13.5 7.9 12.3

7.5 4.5 6.9

0.03 2.3 0.09

3.1 0.0 2.5

22.6 28.4 23.8

14.7 12.5 14.3

10.4 14.8 11.3

17.6 13.6 16.7

6.6 10.2 7.4

5.7 5.7 5.7

2.8 3.4 2.9

2.5 1.1 2.2

6.0 3.4 5.4

Freq = QEEG frequency distribution within normal limits; All - all QEEG measures within normal limits; AMF = alpha mean frequency; Interhem only = only interhemispheric abnormality present.

regions relative to the normal population. Significant hypercoherence represents increased synchronization across cortical regions relative to the normal population. Significant incoherence, both within and between hemispheres, occurred in approximately one third of the normal-IQ and low-IQ children. Incoherence within hemispheres was localized between frontal/occipital regions in 76.9% and between frontal/temporal regions in 23.1%. Incoherence across hemispheres was localized between posterior cortical regions two thirds of the time, occurring more frequently in the low-IQ children. Significant hypercoherence within cortical hemispheres was present in 26.3%, with hypercoherence between hemispheres present in 35.1%. Hypercoherence within hemispheres was localized between frontal/temporal regions 85.9% of the time. Hypercoherence between hemispheres was localized between frontal regions 47.6% of the time, between posterior regions 15.3% of the time, and between both frontal and posterior regions 30.1% of the time. The normal- and low-IQ group averages for all monopolar developmental deviation measures were significandy different from the normal population, with the differences greater for frontal and right hemisphere regions, and with the magnitude of the deviations greater for the low-IQ group. Overall, 31.4% of the normal-IQ and 47.7% of the low-IQ children showed either focal or generalized developmental deviation. Evidence of maturational lag as the source of QEEG abnormality was quite rare, occurring in 6.3% of the normal and 10.2% of the low-IQ children. Developmental deviation was prevalent in frontal and midline regions, whereas maturational lag, if present, was localized within posterior regions.

Neurophysiological Subtypes of Abnormality Because of the preponderance of theta and alpha excess QEEG profiles observed, we examined the nature of these patterns of neurophysiological dysfunction in more detail. Generalized or focal theta/alpha excess was present in 76.2% of our sample of ADD, ADHD, and ATT children. These theta and alpha excess children can be divided into two distinct neurophysiological groups. The most common

group shows evidence of overactive theta and/or alpha generators. This group included 46.4% of our sample, and was composed of those children with theta and/or alpha excess with normal alpha mean frequency. The second group of theta/alpha excess children shows evidence of a slowing of the alpha generator. This group included 29.8% of the total sample and involved theta/alpha excess accompanied by decreased alpha mean frequency.

QEEG and Clinical Classification Separate analyses of variance (ANOVAs) were run on each QEEG measure, to identify differences between the various clinical groups of children. Within the normal-IQ group, 2 (no-LD vs. LD) × 3 (ADHD vs. ADD vs. ATT) level ANOVAs were run on each QEEG measure. Oneway ANOVAs with three levels (ADHD vs. ADD vs. ATT) were calculated on each QEEG measure within the low-IQ group. One-way ANOVAs with two levels were run on each QEEG measure comparing the normal- with the low-IQ groups. The largest QEEG differences were found between the low- and normal-IQ groups. These QEEG differences reflected the degree rather than the type of abnormality found. In frontal regions, the low-IQ group showed greater increases in theta and alpha absolute power, greater increases in alpha hypercoherence, and more extreme decreases in alpha mean frequency than did the normal-IQ children. In midline regions, the low-IQ children showed more pronounced decreases in alpha mean frequency and incoherence than did the normal-IQ children. In posterior regions, the low-IQ children showed increased delta and decreased alpha relative power in comparison to the children with normal IQs. The QEEG abnormalities present within each group reflected deviations from normal development, with the degree of deviation greater in the low-IQ group, especially in frontal regions. Differences in degree and type of QEEG abnormality were noted between the children with and without a learning disorder present. Children with a learning disorder showed more extreme frontal hypercoherence and parietal incoherence than did the children without a

958

BIOLPSYCHIATRY

R.J. Chabot and G. Serfontein

1996;40:951-963

T a b l e 4. P e r c e n t a g e o f A t t e n t i o n D e f i c i t D i s o r d e r C h i l d r e n w i t h S i g n i f i c a n t P o w e r A s y m m e t r y Intrahemispheric

Interhemispheric Clinical subgroup Normal IQ scores (n = 319) Low IQ scores (n = 88) Overall (N = 407)

Negative asymmetry

Positive asymmetry

Negative asymmetry

Positive asymmetry

27.9 38.6 30.2

5.0 4.5 4.9

5.3 12.5 6.9

44.2 55.7 46.7

Negative asymmetry = right hemisphere power greater than Left Hemisphere power or posterior power greater than frontal power (intrahemispheric). Positive asymmetry = left hemisphere power greater than right hemisphere power or posterior power greater than frontal power (intrahemispheric)

learning disorder. The no-LD subgroup showed a greater elevation of alpha relative power in frontal, central, and temporal regions than did the LD subgroup. Qualitative differences were also noted, as the no-LD subgroup was characterized by central and posterior temporal hypercoherence, whereas the LD subgroup showed incoherence within these regions. The LD subgroup also showed some frontal/temporal asymmetry and posterior total power excess not seen in the no-LD subgroup. Differences between the ADHD, ADD, and ATT groups were minimal, reflecting the degree of QEEG abnormality present. Specifically, the ADHD and ADD children showed greater frontal theta excess and frontal hypercoherence than did the children with only attention problems, although all groups showed excess frontal theta and frontal hypercoherence. The children with only attention problems showed slightly elevated frontal beta relative power, whereas the other two groups did not. Parietal incoherence was present in all groups, with the degree of incoherence greater in the ADHD and ADD children. Stepwise discriminant analyses were computed to examine the sensitivity and specificity of the QEEG for classifying these groups of children with attention problems. The QEEG variables entered were selected on the basis of the ANOVAs described above. Discriminant functions, with randomized split-half replications, were calculated comparing: 1) the normal- with the low-IQ children; 2) the ADHD and ADD subgroups within the normal-IQ population; and 3) the children with and without learning disorders.

Normal- versus Low-IQ Children A discriminant function using eight QEEG multivariate features (John et al 1983) resulted in the correct identification of 70.4% of the normal- and 70.5% of low-IQ children. Split-half replication resulted in a sensitivity (low IQ as low IQ) of 59.1% and a specificity of 60.0%, only slightly better than chance accuracy (50%). For each of the variables entered the low-IQ children showed greater average abnormality than the normal-IQ children.

ADHD versus A D D Children A discriminant function using six multivariate and one univariate (P3/P4 alpha asymmetry) QEEG measures resuited in the correct classification of 71.4% of the ADHD and 70.3% of the ADD children. A split-half replication resulted in a sensitivity of 67.1% (ADHD as ADHD) and a specificity of 65.6% (ADD as ADD). When the ATT children with normal IQ scores were run against this discriminant function, 58.8% were classified as ADD.

Learning Disorder versus No-Learning Disorder Children A discriminant function resulting in greater than 60% split-half replication could not be obtained. The best initial discriminant used eight QEEG measures, resulting in correct classification of 64.7% of the no-LD and 70.3% of the LD children.

Table 5. Percentage of Attention Deficit Disorder Children with Significant Coherence Abnormality Interhemispheric Clinical subgroup Normal IQ scores (n = 319) Low IQ scores (n = 88) Overall (n = 407)

Intrahemispheric

Incoherence

Hypercoherence

Incoherence

Hypercoherence

25.4 30.7 26.5

34.8 36.4 35.1

32.6 31.8 32.4

26.6 25.0 26.3

Incoherence = decreased correlation between or within hemispheres relative to normal. Hypercoherence = increased correlation between or within hemispheres relative to normal.

QEEG in Children with Attention Problems

BIOLPSYCHIATRY

959

1996;40:951-963

Discussion The overwhelming majority of children in our sample presenting with attention problems had abnormal eyesclosed resting QEEGs, indicating some degree of neurophysiological dysfunction. Further, these children with attention problems could be discriminated from normal children with high levels of accuracy using a small set of QEEG features. The sensitivity of QEEG was 93.7%, and the specificity was 88%. Given these findings and those of Mann and associates (Mann et al 1992), we disagree strongly with the conclusion of Phillips, who, based upon conventional EEG, concluded that "routine EEG screening is of limited value in childhood behavior disorders" (Phillips et al 1993). Rather, we conclude that although conventional EEG may be of limited value, QEEG can play an important role in documenting cerebral dysfunction in children with attention problems, and is clinically useful in making an initial diagnosis of attention deficit disorder. The patterns of QEEG abnormality observed across children with attention problems were quite homogenous, especially given the range of clinical symptoms displayed by these children. Despite differences in IQ, the presence or absence of a learning disorder, and a wide range of scores on hyperactive, inattentive, and impulsive behavior, QEEG differences across groups were reflected mainly in the degree and not the type of QEEG abnormality present. Thus, children with ADHD showed more generalized and extreme QEEG abnormality than did those with ADD, who showed greater abnormality than the children with only attention problems (ATT). In a similar fashion, children in the low-IQ group were more likely to have a generalized QEEG abnormality, with the incidence of significant asymmetry and incoherence greater in these children than in the normal-IQ children. Further, children with a learning disorder showed greater disturbances in interhemispheric function, whereas those without learning disorders showed greater increases in theta and/or alpha power. Some differences in type of QEEG abnormality were observed across these clinical groups, since the classification accuracy of the discriminant functions comparing the various subtypes of ADD/ADHD were above chance levels; however, the sensitivity and specificity estimates for these discriminant functions were much lower than those obtained for the normal vs. ADD/ ADHD discriminant. This suggests that within the population of children with attention problems, the majority of QEEG variance is accounted for by the normal/abnormal dimension, and not by the dimensions of IQ, or the presence or absence of hyperactivity or learning problems. It is interesting to note that the types

of abnormal QEEG features found in samples of children with specific and generalized learning disorders without ADD/ADHD differ substantially from the abnormal QEEG features characteristic of our sample of ADD/ADHD children (John et al 1983). This indicates that QEEG should prove useful in distinguishing children with ADD/ADHD from those with specific developmental learning disorders, and that both may be distinct neurophysiological entities (Chabot et al 1996). Despite the homogeneity of QEEG abnormality observed across the various groups of ADD, ADHD, and ATT children, there was evidence that different neurophysiological subtypes existed that spanned all clinical groups. We argue that these neurophysiological subtypes result from the interaction of the cortical and subcortical structures contained within the frontal/striatal system, and that these subtypes span the range of attention, hyperactivity, and learning problems present within our population of children. Two primary neurophysiological subtypes were identified. The first subtype occurred in 46.4% of our sample of children, and included those with theta/alpha excess with normal alpha mean frequency. Frontal and/or central regions were most implicated within these children. This QEEG profile has been reported to occur in ADHD (Gittelman et al 1985), depression (Pollock and Schneider 1990), and crack cocaine abuse (Alper et al 1990). Interestingly, increased alpha predicted subsequent positive response to buproprion in children with ADHD (Simeon et al 1986). Perhaps children showing this QEEG profile have overactive theta/alpha generators. The second neurophysiological subtype included 29.8% of the children. These children had theta/alpha excess accompanied by decreased alpha mean frequency. This EEG finding was reported in other studies of children with attention problems (Lubar et al 1985; Lubar 1991; Mann et al 1992; Matsuura et al 1993). Children within this neurophysiological subtype can be characterized along a continuum of EEG slowing occurring across all cortical regions, but more often across frontal regions. Current theories suggest that alpha results from cortical/ cortical interactions driven by a thalamic pacemaker, with theta resulting from either slowed alpha or a hippocampal/ septal pacemaker (Steriade et al 1990). Thus: 1) the first subtype shows increased cortical arousal resulting from increased thalamic alpha generator output and/or a disinhibition of hippocampal theta generators; and 2) the second subtype shows a continuum of EEG slowing resulting from decreased metabolic activity in anterior cortical regions and/or in thalamic or hippocampal pacemakers. The two neurophysiological subtypes described above support the notion that CNS arousal can be abnor-

960

toOL PSYCHIATRY 1996;40:951-963

mally low or high in children with attention problems (Frank 1993). A variant of the first neurophysiological profile may be present in the group of children showing an excess of beta activity. Beta activity is believed to result from cortical/ cortical and thalamo/cortical interactions (Steriade et al 1990), and increased beta may indicate cortical hyperarousal (Morihisa et al 1983). Beta excess has been noted in four of five studies of adult patients with major affective disorders (Pollock and Schneider 1990), and has been associated with anxiety (Kiloh et al 1981). Excess delta was quite rare in our population of children with attention problems, occurring in only 2.9%, but localized to posterior regions 60% of the time. Delta activity is generated between cortical layers II and IV, and may occur as a result of lesions to the thalamus and/or midbrain reticular formation (Steriade et al 1990). Posterior delta excess in hyperactive children has been reported elsewhere (Matsuura et al 1993), although John and associates noted much higher levels of delta excess in children with specific and generalized learning disorders (John et al 1983). Although both generalized and focal patterns of abnormal function were observed, frontal cortical regions were implicated more often than other regions. For example, a generalized theta excess with greater involvement of frontal regions was the most common abnormality found, and in the children with a focal theta excess, this theta was localized to frontal regions 75% of the time. Further, theta and/or alpha hypercoherence was highly prevalent, with the prefrontal area most effected. These findings indicate that frontal lobe dysfunction is quite common in children with attention problems (Benson 1991; Heilman et al 1991). These findings agree with the frontal theta excess found in hyperactive boys by Matsuura and associates (Matsuura et al 1993), but do not support their conclusion of brain immaturity as the biological basis of this abnormal finding. The maturational lag and developmental deviation measures suggest that frontal and midline dysfunction reflects deviations from normal development that cannot be explained as immature EEG patterns. Further, no significant differences in the degree or type of neurophysiological dysfunction were noted as a function of age within our sample of 6-17-year-old children. Approximately one third of our sample of children with attention problems showed signs of disturbed interhemispheric function indicative of decreased coordination of neurophysiological activity between hemispheres. Interhemispheric power asymmetry was quite common, most often localized between left and right posterior regions, with excess fight hemisphere power eight times as likely

R.J. Chabot and G. Serfontein

to occur as excess left hemisphere power. Significant interhemispheric incoherence was also quite prevalent, especially in posterior and midline regions, indicating abnormal communication between these cortical regions in children with attention problems. The interhemispheric findings, taken in conjunction with the measures of developmental deviation, indicate that abnormal fight hemisphere function is highly prevalent in ADD/ADHD, especially in posterior temporal and parietal regions (Heilman et al 1991), and is often accompanied by dysfunction in interhemispheric communication. These abnormal QEEG findings reflect disturbances in corpus callosum functioning and highlight the importance of normal interhemispheric relationships to attention and learning processes. Disturbances in intrahemispheric function were also quite common, especially between frontal and posterior regions. This intrahemispheric incoherence indicates disturbed cortical/cortical relationships that are modulated by subcortical interconnections via the thalamus and/or basal ganglia (Newton et al 1993). The excess theta and alpha power described above was often accompanied by hypercoherence within these frequency bands. This indicates the presence of hypersynchronous EEG activity in the resting EEGs of children with attention problems. The localization of such hypersynchronous activity between left and right frontal regions (especially frontal/polar regions), as well as the high incidence of hypercoherence within frontal/temporal intrahemispheric regions, may indicate a lack of cortical differentiation within these brain regions so important for maintaining and directing attention and learning processes. In normal children frontal coherence decreases with age, but this decrease does not occur in children with learning disorders (Marosi et al 1992), and hypercoherence remained highly prevalent across all ages within our sample of children. Hypercoherence between closely paired cortical regions most likely reflects hypersynchronous intracortical activity (Newton et al 1993). These results replicate most EEG findings reported in children with ADD/ADHD, help clarify the disparate findings reported, and suggest neurophysiological dysfunction in those cortical and subcortical anatomical structures implicated by positron emission tomography, MRI, and regional cerebral blood flow studies. In particular, frontal/striatal and corpus callosum dysfunction and/or structural abnormality are often present in ADD/ADHD. Zametkin and associates (Zametkin et al 1990, 1993) report decreased glucose metabolism in left anterior frontal regions, in premotor and motor cortex (frontal and central regions), in thalamic regions, in the hippocampus, and in right temporal and posterior/temporal regions of teenagers and adults with ADHD. Many of these same regions have been reported to show structural MRI abnor-

QEEG in Children with Attention Problems

BIOLPSYCHIATRY

961

1996;40:951-963

reality as well in A D D / A D H D (Castellanos et al 1994; Giedd et al 1994; Hynd et al 1990, 1991). Abnormal regional cerebral blood flow has also been reported in hyperactive children, with hypoperfusion within the right striatal region and hyperperfusion within sensory and sensory/motor cortex (possibly due to thalamic inhibition) (Lou et al 1989). These findings suggest that attention deficit disorder may result from a developmental abnormality in the frontal/cortical, striatal, and thalamic circuits with right hemisphere dysfunction also indicated. The QEEG and imaging data implicate frontal lobe and right hemisphere dysfunction, with disturbances both within and between hemispheres present in children with ADD/ ADHD. The abnormal QEEG profiles described above highlight the relationship between EEG slowing and dysfunction in attention and cognition noted in other neurological and psychiatric patient populations (Brenner et al 1986; Coben et al 1985; John and Prichep 1990; Ritchlin et al 1992). The high incidence of both interhemisphefic and intrahemispheric abnormality indicates the importance of cortical/cortical and cortical/subcortical interactions both

within and across hemispheres for normal attention and learning processes. The abnormal QEEG profiles noted across our population of children were quite homogenous, despite the heterogenous nature of the clinical symptoms of inattention, impulsivity, hyperactivity, and learning problems observed. We have hypothesized that the various QEEG profiles identified represent two to three basic neurophysiological subtypes that span the clinical symptom space defined in our A D D / A D H D / A T T population. We will attempt to show the relationship between these subtypes and treatment response to various pharmacologic agents in future research.

The authors wish to thank the staff of the Serfontein clinic, and in particular Lisa Wood, for their diligent work in collecting all data. We also thank Henry Merkin of the Brain Research Labs for his effort in processing and inspecting all EEG data prior to quantification. We also thank Dr. E. Roy John for his comments, which greatly helped increase the clinical relevance of this paper, and Dr. Leslie Prichep for her patience and help in understanding the implications of our QEEG findings.

References Alper KR, Chabot RJ, Kim AH, Prichep LS, John ER (1990): Quantitative EEG correlates of crack cocaine dependence. Psychiatry Res 35:95-106. Alvarez A, Pascual R, Valdes P (1987): U.S. EEG developmental equations confirmed for Cuban schoolchildren. Electroencephalogr Clin Neurophysiol 67:330-332. Alvarez A, Sosa V, Marqui P, Garcia G, Biscay L, Bayard B (1989): On the structure of EEG development. Electroencephalogr Clin Neurophysiol 73:10-19. August G, Garfinkel B (1989): Behavioral and cognitive subtypes of ADHD. J Am Acad Child Adolesc Psychiatry. 28:739-748. Benson D (1991): The role of frontal dysfunction in attention deficit hyperactivity disorder. J Child Neurol 6(suppl):S9S12. Brenner RP, Ulrich RF, Spiker DG, et al (1986): Computerized EEG spectral analysis in elderly normal, demented and depressed subjects. Electroencephalogr Clin Neurophysiol 64:483-492. Callaway E, Halliday R, Naylor H (1983): Hyperactive children's event-related potentials fail to support underarousal and maturational-lag theories. Arch Gen Psychiatry 40:12431248. Castellanos FX, Giedd JN, Eckburg P, et al (1994): Quantitative morphology of the caudate nucleus in attention deficit hyperactivity disorder. Am J Psychiatry l 51:1791-1796. Chabot R, Merkin H, Wood L, Davenport T, Serfontein G

(1996): Sensitivity and specificity of QEEG in children with attention deficit or specific developmental learning disorders. Clin Electroencephalogr 27:26-34. Coben LA, Danziger W, Storandt M (1985): A longitudinal EEG study of mild senile dementia of Alzheimer type: Changes at 1 year and at 2.5 years. Electroencephalogr Clin Neurophysiol 61:101-112. Dykman RA, Holcomb PJ, Oglesby DM, Ackerman PT (1982): Electrocortical frequencies in hyperactive, learning-disabled, mixed, and normal children. Biol Psychiatry 17:675-685. Frank Y (1993): Visual event related potentials after methylphenidate and sodium valproate in children with attention deficit hyperactivity disorder. Clin Electroencephalogr 24: 19-24. Gasser T, Mochs J, Lenard HG, Bacher P, Verleger R (1983): The EEG of mildly retarded children: Developmental, classificatory and topographic aspects. Electroencephalogr Clin Neurophysiol 55:131-144. Giedd J, Castellanos F, Casey B, et al (1994): Quantitative morphology of the corpus callosum in attention deficit hyperactivity disorder. Am J Psychiatry 151:665-669. Gittelman R, Mannuzza S, Shenker R, Bonagura N (1985): Hyperactive boys almost grown up. Arch Gen Psychiatry 42:937-947. Harmony T (l 988): Psychophysiological evaluation of children' s neuropsychological disorders. In Reynolds CR (ed), Handbook of Child Clinical Neuropsychology. New York: Plenum, chapter 15, pp 265-290.

962

BIOL PSYCHIATRY 1996;40:951-963

R.J. Chabot and G. Serfontein

Heilman K, Voeller K, Nadeau S (1991): A possible pathophysiologic substrate of attention deficit hyperactivity disorder. J Child Neurol 6(suppl):S74-S79.

Shabsin HS (1985): Spectral analysis of EEG differences between children with and without learning disabilities. J Learning Disabilities 18:403-408.

Hynd G, Semrud-Clikeman M, Loryus A, Novey E, Eliopulos D (1990): Brain morphology in developmental dyslexia and attention deficit disorder/hyperactivity. Arch Neurol 47:919926.

Mann CA, Lubar J, Zimmerman A, Miller C, Muenchen R (1992): Quantitative analysis of EEG in boys with attention deficit hyperactivity disorder: Controlled study with clinical implications. Pediatr Neurolo 8:30-36.

Hynd G, Semrud-Clikeman M, Lorys A, Novey E, Eliopulos D, Lyytinen H (1991): Corpus callosum morphology in attention deficit-hyperactivity disorder: Morphometric analysis of MRI. J Learning Disord 24:141-146.

Marosi E, Harmony T, Sanchez L, et al (1992): Maturation of the coherence of EEG activity in normal and learning-disabled children. Electroencephalogr Clin Neurophysiol 83:350357.

John ER, Prichep LS (1990): Neurometric studies of aging and cognitive impairment. In Uylings HBM, Eden CGV, Bruin JPCD, Corner MA, Feenstra MGP (eds), The Prefrontal Cortex, ITS Structure, Function and Pathology. Progress in Brain Research. Amsterdam: Elsevier, pp 545-555.

Matsuura M, Okubo Y, Toru M, et al (1993): A cross-national study of children with emotional and behavioral problems: A WHO collaborative study in the Western Pacific region. Biol Psychiatry 34:59-65.

John ER, Ahn H, Prichep LS, Trepetin M, Brown D, Kaye H (1980): Developmental equations for the electroencephalogram. Science 210:1255-1258. John ER, Prichep LS, Ahn H, Easton P, Fridman J, Kaye H (1983): Neurometric evaluation of cognitive dysfunctions and neurological disorders in children. Prog Neurobiol. 21:239290. John ER, Prichep LS, Easton P (1987): Normative data banks and neurometrics: Basic concepts, methods and results of norm construction. In Remond A (ed), Handbook of Electroencephalography and Clinical Neurophysiology, Vol II1, Computer Analysis of the EEG and Other Neurophysiological Signals. Amsterdam: Elsevier, pp 449-495. John ER, Prichep LS, Fridman J, Easton P (1988): Neurometrics: Computer assisted differential diagnosis of brain dysfunctions. Science 293:162-169. John ER, Prichep LS, Ahn H et al (1989). Neurometric Evaluation of Brain Function in Normal and Learning Disabled Children. Ann Arbor, MI: University of Michigan Press. John ER, Prichep LS, Almas M (1992): Subtyping of psychiatric patients by cluster analysis of QEEG. Brain Topogr 4:321326. Jonkman EJ, Poortvliet DCJ, Veering MM, deWeerd AW, John ER (1985): The use of neurometrics in the study of patients with cerebral ischemia. Electroencephalogr Clin Neurophysiol 61:333-341. Kaye H, John ER, Ahn H, Prichep LS (1981): Neurometric evaluation of learning disabled children. Int J Neurosci 13:15-25. Kiloh LG, McComas AJ, Osselton JW, Upton ARM (1981): Clinical Electroencephalography. London: Butterworths. Lopes da Silva FH (1990): A critical review of clinical applications of topographic mapping of brain potentials. J Clin Neurophysiol 7:535-55 I. Lou H, Henriksen L, Bruhn P, Borner H, Nielsen J (1989): Striatal dysfunction in attention deficit and hyperkinetic disorder. Arch Neurol 46:48-52. Lubar J (1991): Discourse on the development of EEG diagnostics and biofeedback for attention-deficit/hyperactivity disorders. Biofeedback Self-Regul 16:201-224. Lubar JF, Bianchini KJ, Calhoun WH, Lambert EW, Brody ZH,

Montagu JD (1975): The hyperkinetic child: A behavioural, electrodermal and EEG investigation. Develop Med Child Neurol 17:299-305. Morihisa JM, Duffy FH, Wyatt RJ (1983): Brain electrical activity mapping (BEAM) in schizophrenic patients. Arch Gen Psychiatry 40:719-728. Newton TF, Leuchter AF, Walter DO, et al (1993): Quantitative EEG in patients with AIDS and asymptomatic HIV infection. J Neuropsychiatry Clin Neurosci 5:316-321. Oken BS, Chiappa KH, Salinsky M (1989): Computerized EEG frequency analysis. Neurology 39:1281-1287. Phillips B, Drake M, Hietter S, Andrews J, Bogner J (1993): Electroencephalography in childhood conduct and behavior disorders. Clin Electroencephalogr 24:25-30. Pollock VE, Schneider LS (1990): Quantitative, waking EEG research on depression. Biol Psychiatry 27:757-780. Prichep LS, John ER (1992): QEEG profiles of psychiatric disorders. Brain Topogr 4:249-257. Ritchlin CT, Chabot RJ, Alper K, et al (1992): Quantitative electroencephalography: A new approach to the diagnosis of cerebral dysfunction in systemic lupus erythematosus. Arthritis Rheumatism 35:1330-1342. Schachar R (1991): Childhood hyperactivity. J Child Psychol Psychiatry 32:155-191. Semrud-Clikeman M, Biederman J, Sprich-Buckminster S, Lehman B, Faraone S, Norman D (1992): Comorbidity between ADHD and learning disability: A review and report in a clinically referred sample. J Am Acad Child Adolesc Psychiatry 31:439-448. Simeon JG, Ferguson HB, Fleet JVW (1986): Bupropion effects in attention deficit and conduct disorders. Can J Psychiatry 31:581-585. Small J (1993): Psychiatric disorders and EEG. In Niedermeyer E, Silva FLD (eds), Electroencephalography: Basic Principles, Clinical Applications, and Related Fields. Baltimore: Williams and Wilkins, chapter 31, pp 581-596. Steriade M, Gloor P, Llinas RR, Lopes da Silva F, Mesulam MM (1990): Report of the IFCN committee on basic mechanisms. Electroencephalogr Clin Neurophysiol 76:481-508. Veldhuizen RJ, Jonkman EJ, Poortvliet DCJ (1993): Sex differences in age regression parameters of healthy adults--Nor-

QEEG in Children with Attention Problems

mative data and practical implications. Electroencephalogr Clin Neurophysiol 86:377-384. Weinberg W, Brumback R (1992): The myth of attention deficit-hyperactivity disorder: Symptoms resulting from multiple causes. J Child Neurol 7:431-443. Weiner JM, Dunn OJ (1966): Elimination of variates in linear discrimination problems. Biometrics 22:268-275. Yingling CD, Galin D, Fein G, Peltzman D, Davenport L (1986):

BIOL PSYCHIATRY 1996;40:951-963

963

Neurometrics does not detect "pure" dyslexics. Electroencephalogr Clin Neurophysiol 63:426-430. Zametkin A, Nordahl T, Gross M, et al (1990): Cerebral glucose metabolism in adults with hyperactivity of childhood onset. N Engl J Med 323:1361-1366. Zametkin AJ, Liebenauer LL, Fitzgerald GA, et al (1993): Brain metabolism in teenagers with attention-deficit hyperactivity disorder. Arch Gen Psychiatry 50:333-340.