A Brain-Wave Test for ADHD, Cleared but Not Endorsed

What the advisory found

In 2013 the US Food and Drug Administration cleared a device called the Neuropsychiatric EEG-Based ADHD Assessment Aid, marketed as NEBA, for patients aged 6 to 17. It records electrical activity at a single point on the scalp and calculates the ratio between the power of two frequency bands, theta and beta. The clearance described the device as confirmatory support for a completed clinical evaluation, and stated that it was not to be used alone in evaluation or diagnosis.

In November 2016 the American Academy of Neurology published its own assessment of that measurement in the journal Neurology, written by David Gloss, Jay Varma, Tamara Pringsheim and Marc Nuwer for the academy’s guideline subcommittee. A practice advisory generates no new data. A panel grades the existing literature by class of evidence and issues a recommendation whose letter records how much confidence that grading will bear.

The search covered three databases without date limits and returned 959 abstracts, each screened independently by two panel members. Thirty-two articles addressed the accuracy question and two survived as Class I, the highest rating. Both were prospective blinded cohort studies enrolling children referred with suspected ADHD rather than children already known to have it, which let the test be judged against the cases it would meet in practice.

In the smaller study, reported by Humberto Quintana and colleagues in 2007, 25 of 26 children and adolescents were classified in agreement with a psychiatric evaluation, and every ADHD identification the EEG made rested on the theta/beta ratio rather than on frontal beta power. The larger, reported by Steven Snyder and colleagues in 2008, evaluated 159 of 160 enrolled participants and reached an overall accuracy of 89 percent. From that figure the panel calculated that 5 of the 62 participants without a clinical ADHD diagnosis had been given one by the EEG, a false-positive rate of 8 percent.

Combined, the two studies identified 166 of 185 participants correctly and misidentified 19. The panel measured that against the estimated 5 percent rate of diagnostic error in US outpatient care and treated anything above it as unacceptable, because a false positive means a child carrying a diagnosis, and often a treatment, for a condition they do not have. Sensitivity and specificity were relatively high, and the combination was still judged insufficiently accurate.

Two recommendations followed. At Level B, clinicians should inform families that the ratio and frontal beta power together should not replace a standard clinical evaluation, and that the false-positive rate carries a risk of significant harm. At Level R, the academy’s designation for procedures belonging in research settings, clinicians should inform families that the ratio should not be used to confirm an ADHD diagnosis or to justify further testing after an evaluation.

What it means when the test is offered

Theta is a highly nonspecific feature of an EEG. Drowsiness raises it, medication raises it, and so do many neurological disorders, while frontal beta power varies between individuals for reasons unconnected to attention. Both measures move with a person’s state of alertness, so a recording taken minutes after another can disagree with it.

The gap the advisory returns to is the differential diagnosis. The question a parent brings to an evaluation is more often which disorder is present than whether one particular disorder is, and depression, anxiety, learning disabilities and oppositional defiant disorder all occupy that field. Nobody has established what theta/beta ratios look like in those conditions, so an elevated ratio cannot separate ADHD from what resembles it.

Children diagnosed with ADHD do show high theta/beta power at that electrode, and the advisory ends by asking for research into why, along with work on test-retest reliability and medication effects. What the panel rejected was the measurement as a substitute for the examination. Whether it adds anything to a completed examination rested on a single lower-graded study, and the panel recorded that question as unanswered.

That examination is substantial in its own right. The American Academy of Pediatrics asks that DSM criteria be met with impairment documented in more than one major setting, and that information be gathered from parents, teachers and the young person. The parent-completed Conners 3 ADHD Index, one of the rating scales used in that process, discriminates children with ADHD from non-clinical children with a sensitivity of 92 percent and a specificity of 94 percent.

A clinic that offers EEG-based assessment can be asked what the result will be used for. An answer placing the recording after a completed clinical evaluation, as one input among others, matches the terms under which the device was cleared; an answer offering the recording as the thing that will establish whether a child has ADHD describes a use the clearance excluded and the neurologists confined to research. The advisory reflects the evidence available when it was written, and the academy notes that its guidance is not continually updated, so the two Class I studies are the strongest evidence the panel had rather than the strongest that exists.

If a recording is already in your child’s file, the clinical evaluation remains the document that carries the diagnosis.