Limitations of the American Academy of Neurology and American Clinical Neurophysiology Society Paper on QEEG

One example is the AAN/ACNS 1997 position regarding traumatic brain injury, which is given a Type D recommendation. Although the AAN/ACNS report “does not attempt to cite all QEEG literature,” nevertheless, the article omits reference to several Class II studies that appear to meet the standards for Type B recommendation.^3–8 If these additional studies had been presented, then readers of the AAN/ACNS report might have drawn a different conclusion. Further, it is difficult to understand why a judgment of “inconclusive” evidence is rendered for QEEG and brain injury when greater than 95% sensitivity and 89% specificity of the QEEG has been reported in publications in refereed journals.^9,10 The level of sensitivity and specificity of QEEG for TBI surely meets the clinical standards maintained for MRI, sonograms, blood analysis, and other common clinical diagnostic measures. The published specificity and sensitivity of QEEG in traumatic brain injury⁹ meets the standards of sensitivity and specificity enumerated by the AAN/ACNS paper, yet it is still placed in the “rejected” category.

The AAN/ACNS paper also inaccurately reports conflicting Class II evidence in QEEG and brain injury. For example, the Type D recommendation in the AAN/ACNS paper relied on the following: 1) a QEEG study of mild traumatic brain injury by Tebano et al.,¹¹ in which Nuwer stated: “In one small group of patients with postconcussion syndrome, an increase in 8 to 10 Hz alpha was reported. A subsequent report described reduced alpha in a much larger group of patients after mild head injury” (p. 283), and 2) the assertion that “Others have commented that this technique is predisposed to false-positive abnormalities in normal subjects due to mild drowsiness or other problems” (p. 283).

The “subsequent report” referred to above was a study by Thatcher et al.⁹ The AAN/ACNS's juxtaposition of the italicized words “increase” and “reduced” alpha implies opposite findings between the study by Tebano et al.¹¹ and the study by Thatcher et al.⁹ when, in fact, there is no discrepancy. For example, Tebano et al.¹¹ also reported a shift toward lower alpha frequencies as well as reduced 10.5- to 13.5-Hz alpha and reduced beta frequency EEG amplitudes, which is very similar to the findings reported by Thatcher et al.^9,10

In addition, the paper misleads because “others” were not identified and there were no citations by AAN/ACNS of scientific evidence that refutes or contradicts the findings of Thatcher et al.^9,10 or Tebano et al.¹¹ In fact, the AAN/ACNS paper referred to Thatcher et al.^9,10 by confirming that the authors “were able to replicate their findings with good sensitivity and specificity” (p. 283). It would appear that the Nuwer paper contradicted itself and arbitrarily discounted, without scientific justification and only by reference to anonymous “others,” at least three well-controlled studies, including one study that involved 608 mild TBI patients and 108 age-matched control subjects with independent cross-validations.⁹

Therefore, one limitation of the position taken by AAN/ACNS is that it is not scientifically balanced, nor does it accurately state the value of QEEG in the detection of TBI. These facts, as well as recent correlations between QEEG, MRI, and neuropsychological performance in military and veteran TBI patients from the multicenter Defense and Veterans Head Injury Program, emphasize the need for a reevaluation of the AAN/ACNS's position on this matter.^12–14 QEEGs are performed at baseline, 12 months, and 24 months into the rehabilitation program for TBI in four major Veterans Affairs hospitals (Palo Alto, Tampa, Richmond, and Minneapolis) as well as three major military bases (Balboa Naval Medical Center, Wilford Hall Air Force Hospital, and Walter Reed Army Medical Center).

The 1997 AAN/ACNS report emphasizes the problem of false positives because of the large number of statistical tests, but the report also fails to cite the many studies in which test-retest reliability and independent cross-validation were used.^9,15,16 The AAN/ACNS report also omitted reference to the clinically established sensitivity and specificity of observed lifespan QEEG reference databases to describe and predict the connectivities (coherence), conduction delay times (phase), and excitabilities (amplitude asymmetries, spectral power) among and within the cerebral systems represented by the standard 10/20 montage.^15,17–22 While the QEEG is but one diagnostic test among many that may assist a health professional in formulating a clinical judgment, it nonetheless merits a fair assessment of its usefulness. In the 1997 report, the use of normative databases, cross-validation, test-retest reliability and other such procedures were not discussed or fairly presented.

Another example of bias and misrepresentation is the undocumented assertion that “drowsiness” reduces the accuracy of a QEEG head injury discriminant function (p. 278). Drowsiness is an artifact that can be easily eliminated and prevented.¹² The elimination of artifact is important in all of diagnostic medicine, including all the “gold standard” techniques of diagnosis. Bias is evident in the elevation of the importance of artifact elimination for QEEG above and beyond that which is expected in all areas of science and diagnostic medicine.

The research cited in the AAN/ACNS paper, while thorough in some respects, leaves out significant contributions in areas that we believe support uses different from standard electroencephalography. The paper therefore artificially creates an impression that the use of QEEG is limited. High-quality peer-reviewed papers attesting to an expanding list of uses of QEEG appear with increasing frequency in the scientific and clinical literature. Using normal subjects as their own controls, it has been possible to show task-specific, localized, treatment-specific changes in the QEEG. For example, the benefit for patients with intractable seizures^23–26 could be obscured by a literal interpretation of the AAN/ACNS paper. QEEG provides an invaluable assessment and outcome measurement tool for the application of EEG biofeedback (commonly referred to as neurofeedback) therapy for seizure disorders. This method uses basic principles of learning and contemporary neurophysiology, together with advanced computer technologies, to teach patients to alter deviant EEG patterns and thereby reduce clinical symptoms. Thus, many epileptic patients have been able to reduce seizures, reduce or withdraw from medications, or avoid neurosurgery through this treatment approach.^23,24,27,28 The clinical utility and importance of QEEG in the field of neurofeedback can be seen, for example, in evaluations of the course of therapy. One can use a reference EEG database to evaluate the location and type of EEG feature to target for neurofeedback training (i.e., locations with EEG abnormalities greater than two standard deviations from a normative group).²⁹

With the quantitative EEG and topographic brain maps, it is often possible to observe attributes of brain function that cannot be seen in the raw EEG signal. One example is in the evaluation of cognitive processes in normal individuals. These processes can be observed and quantified through subtle frequency-related and coherence-related activities in the QEEG brain maps that index the degree of difficulty of cognitive tasks. Measurements of the suppression or enhancement of components within the topographic EEG through the averaging of responses in the frequency domain at time intervals surrounding a point of cognitive or sensory stimulation have been described.^30,31 These measures provide a means for the reliable examination of the timing, degree, and functional specificity of cortical responses to cognitive events. Recent studies have shown that this method can be used to objectively evaluate the cortical responses and physiological substrates of most cognitive functions, including attention, perception, memory, and psychomotor performance.^32–35 Indeed, given the temporal and neurophysiological resolution that this method provides, it may truly alter the way clinical EEG data are collected and evaluated in the future.

We agree that it is essential that the unprocessed analog EEG be recorded and reviewed during and after the collection of data in quantitative EEG and brain mapping. Analog data disclose artifact, and their review ensures that valid data contribute to the primary conclusions of quantitative evaluations. Direct examination of the raw signal also addresses any questions that may arise regarding the cross-validation between quantitative and analog EEG analysis. These practices are fundamental to the responsible application of QEEG methodology and are accepted doctrine within this field.

With regard to diagnosis and treatment of mild traumatic brain injury, there are rehabilitation treatments based on neuropsychological testing and the correlation of the test findings with QEEG findings. One of the advantages of QEEG evaluation in working with closed head injury is that it provides localization for specific areas of the cortex where coherence and frequency relationships are abnormal in comparison with a normative database.^{9–11,13,15,29,36,37} Such comparisons cannot be done by simply looking at the analog EEG, no matter how skilled one is at visual pattern recognition. Once comparative information is obtained with quantitative EEG techniques and correlated with history and neuropsychological findings, EEG neurofeedback and other treatments can be administered.^29,38–40

Another area in which there has been considerable progress based on QEEG correlates is the evaluation and treatment of attention-deficit/hyperactivity disorder (ADHD).^41–45 Papers authored by several researchers^{15,17,41,42,46–49} on the subject of attention deficit disorder (ADD) were mentioned in only one context, yet research demonstrates the strong correlates of QEEG with ADHD, as well as the clear utility of QEEG in helping with diagnoses and predicting/evaluating treatment response. Although there are analog EEG studies from the 1960s and 1970s that were suggestive of some EEG markers associated with ADD/ADHD, a recent QEEG study⁵⁰ showed that there was a significant excess in the amount of slow activity in boys with the inattentive form of ADD/ADHD, in comparison to normal control subjects. Moreover, the excess slow activity occurred primarily during academic challenges such as reading or drawing. One of the problems in the past was that many EEG analyses were done only under baseline conditions, that is, eyes open or eyes closed. However, important discriminative differences may occur only when cognitive challenges are introduced. Many of these changes are subtle, involving phase, coherence, and frequency relationships that cannot be seen in a raw analog EEG evaluation.^{15,17,41,42,51}

Journals specializing in other areas contain valuable publications and data pertinent to QEEG that were not included. Although it would be impossible for the AAN/ACNS team to have included everything, the paper leaves the impression that these other areas are not valid because they were not sanctioned by the AAN/ACNS team. That is, the studies cited above are examples of Class I and Class II studies with reasonable clinical sensitivity and specificity. Thus, there is solid scientific research supporting the use of QEEG with these disorders. This is not the case for many of the psychiatric diagnoses and treatments accepted by AAN/ACNS.

We do agree that QEEG, no matter how thorough, should not be the only tool used for diagnosis. For example, in the area of attention deficit disorder there is no single technique that can be solely relied upon for the diagnosis. This is due in part to the vagaries of this diagnosis, despite its roots in neurologic disturbance. Manifestations of ADD/ADHD reflect behavior problems,^52,53 learning style, cognitive processing, social interaction, and many other developmental factors. The current diagnosis of ADD/ADHD depends also on the use of computerized continuous performance tasks, detailed history, school performance, and evaluation for learning disabilities and other comorbidities, as well as other measures. QEEG data complement these other findings by providing for a comparison of brain activity with databases for both normal and ADD/ADHD groups.

We likewise propose multidimensional analysis in diagnosis of other conditions, but it must be remembered that there are many conditions that are neurologically based, such as learning disabilities. Even where there are PET scan findings, and in some cases evidence of cortical dysplasia,⁵⁴ QEEG may give us more information than analog EEG. Often, neurologically based disorders do not involve a structural abnormality, lesion, or disease process, but abnormalities are expressed in the way the brain evaluates information. These processes can be studied with QEEG techniques, but not with simple visual analysis of the unprocessed analog signal.

If the use of QEEG techniques were eliminated or reduced, the clock of science would be turned back to the 1950s, when EEG technologists and EEG experts used calipers to measure frequencies and EEG analyses often took hours instead of minutes to perform. We strongly believe that the area of QEEG and topographic brain mapping should be actively pursued and developed. There should be no restriction concerning the type of professional permitted to use these techniques as long as the person is trained and qualified. We agree with the assertion made by Ernest Rodin that EEG topography “is an exciting field with rapid growth, but it needs to remain in well-trained, competent hands…” and that “both neurologists and non-neurologists who use quantitative EEG need to be trained and competent in its inherent complexities.”⁵⁵ In this regard, one of the most controversial statements in the AAN/ACNS report is the statement that “EEG brain mapping and other advanced QEEG techniques should be used only by physicians highly skilled in clinical EEG, and only as an adjunct to and in conjunction with traditional EEG interpretation.” We strongly believe that adequately trained nonphysicians should also be encouraged and enabled to use these techniques. It is well known that many QEEG methods, procedures, and published studies are based on the efforts of non-neurologists. The most extensive and accurate descriptions of normal human QEEG (cerebral) maturation arose from the work of neuroscientists.^{15,17,21,56–59} Many nonphysicians were responsible for the development of these techniques and are highly trained in the mathematics and physics of signal analysis fundamental to QEEG. In fact, these individuals are often the very people who teach physicians about the use of EEG and QEEG.

Certification programs are currently being developed, or are already in effect, in the areas of conventional electroencephalography (EEG certification), quantitative EEG (for example, certification by the EEG Clinical Neuroscience Society), and biofeedback/neurofeedback (for example, EEG certification by the Biofeedback Certification Institute of America). Along with established certification programs such as that in polysomnography (which also utilizes EEG techniques), these ensure that individuals who use EEG are adequately trained and have demonstrated their competence by passing written and/or practical skill-based examinations as well as completing coursework and experience requirements. These programs provide reasonable safeguards against the unqualified use of this method. Although they can be improved, they clearly would not benefit from the additional requirement of a medical degree.