Neurofeedback, also known as “EEG biofeedback,” has been around for a long time, but its history is checkered. Clinicians have hawked devices for treating ADHD and other psychiatric conditions in the absence of reliable efficacy data. Since the treatment is rarely covered by insurance companies, families may pay several thousand dollars for a typical treatment course.

Recently, however, a new metaanalysis was published and concluded that some recent well-designed studies endorse the technique for ADHD. Let’s go through some of this literature and decide whether neurofeedback is ready for prime time.

First, a little background. Neurofeedback is based on the notion that patients with ADHD have characteristic patterns on the electroencephalogram (EEG). Specifically,some studies have shown that, when presented with a task requiring attention, ADHD patients don’t generate enough fast beta waves (typically associated with focus), and instead generate too many slow theta waves (usually associated with daydreaming) (Fox DJ et al, Appl Psychophysiol Biofeedback 2005;30(4):365–373).

In neurofeedback for ADHD, kids are trained to produce more beta waves and fewer theta waves. A typical session lasts about an hour. The patient is wired up to an EEG monitor, and sits in front of a screen. Computer software translates the EEG waves into video games. In one system, for example, beta waves make a rocket ship fly faster, while theta waves cause a competing rocket to overtake it. The child is rewarded with points for activating beta waves. The sessions are usually weekly, and a treatment course is from 20 to 40 weeks, with each treatment costing about $100. You can watch a short informational video at http://bit.ly/f0hi7k.

The Evidence for Neurofeedback
Until recently, most of the scientific evidence for neurofeedback consisted of either case reports or uncontrolled clinical trials. While many of these trials reported positive results, it was impossible to rule out the placebo effect as the operative mechanism.

The first, and thus far the only large randomized controlled study was published in 2009. In this study, 102 German children with ADHD (ages eight through 12) were randomly assigned to either neurofeedback training (NF) or to computerized attention skills training (AST). The computerized AST consisted of video games to practice attention, vigilance, reactivity, and visual and auditory perception. Researchers considered this a kind of placebo condition because it controls for the nonspecific effects of concentrating on a task. Neurofeedback is thought to be a more specific treatment than simply training kids in attention and vigilance—rather, it presumably teaches them to identify specific brain waves associated with being alert and focused. In order to control for positive expectancy, parents were told that both the neurofeedback and the AST were “experimental but promising treatments for ADHD.”

The participants were not on medication for attention and most had never been on medication, but had presented for outpatient treatment of ADHD. Comorbid tics, dyslexia, emotional disorders, and conduct disorders were allowed, but other comorbidities were not. The children underwent 36 sessions of NF or AST at a frequency of two to three sessions per week. The main outcome measures was the total parent rating on the German ADHD rating scale; secondary outcomes included teacher ratings and response rates, defined as at least a 25% improvement in the rating scale.

Ninety-four children were included in the final analysis. Neurofeedback produced significantly more improvement in parent ADHD ratings than the control condition, with an effect size of 0.6. (This is generally considered a “moderate” effect size. For comparison, note that the effect size of atomoxetine (Strattera) is around 0.71 (Michaelson et al, Am J Psychiatry 2002;159:1896–1901). Effect size is a statistical method that can be used to compare different studies with different methodologies. Here, these are expressed as a Cohen’s d, which is the difference between two means divided by a standard deviation for the data.) Children in the NF group had a response rate of 51.7% vs. 28.6% for the control group, which was also statistically significant. The study itself was sponsored by the German Research Foundation (Gevensleben H et al, J Child Psychol Psychiatry 2009;50(7):780–789).

Neurofeedback has also been evaluated in several smaller studies with generally positive findings, but none as large and as well controlled as this one. If you’re interested in poring over these other studies, you can find a review of them in Arns M et al, Clin EEG Neurosci 2009;40(3):180–189. This was a metaanalysis of all the research neurofeedback for ADHD done to date. It concluded with the rather bullish statement that “we conclude that neurofeedback treatment for ADHD can be considered ‘Efficacious and Specific’ (Level 5) with a large Effect Size for inattention and impulsivity and a medium Effect Size for hyperactivity.”

Caveat emptor, however: the authors are all affiliated with a large Dutch clinic specializing in neurofeedback, so their opinions might be just a teeny bit biased. Nonetheless, overall the research thus far suggests that neurofeedback may indeed be an effective non-drug option for children with ADHD, and that the beneficial effects may be sustained for at least six months after treatment is discontinued. It’s likely that many practitioners would use NF in conjunction with medication, perhaps as a way of weaning kids off stimulants or reducing the dose.

Quantitative EEG Neurofeedback
Quantitative EEG scanning, or qEEG, is another system that has been tested for neurofeedback in ADHD. The simplified hypothesis behind qEEG is this: a database of EEGs is collected and normed—those closer to the average are considered normal, and those on the extremes are compared with other evidence of diagnosis or impairment, all in an attempt to find reliable markers for disease states.

In neurofeedback, the qEEG scan is used to provide the basis of feedback: the participant’s job is to “normalize” his or her brainwaves to look more like the average. This is different from standard EEG neurofeedback, in which patients are trying to enhance specific brain waves. The published research on qEEG neurofeedback is minimal and of poor methodologic quality. The largest study, published in 2002, was an open label trial in which 100 children with ADHD, ages six to 19, were assigned to either qEEG based neurofeedback training plus methylphenidate, or to methylphenidate alone. Both groups also received 10 sessions of parent coaching. The treatment assignments were not done randomly; instead, parents were asked to choose which group they preferred.

The neurofeedback group had weekly EEG sessions which continued until the participant was able to maintain a state in which his or her cortical activity was within one standard deviation of normed peers for a period of 40 minutes on three consecutive trials (an average of 43 sessions were required). The participants were tested at completion of the study and one year later, by parent and by teacher questionnaires as well as by the test of variables of attention (TOVA).

Neurofeedback was found to have a significant effect on improvement in primary symptoms, both on initial and follow up testing, while having taken medication was found to have no effect on the TOVA at one year follow up, no particular surprise. It must be noted as well that the study’s main author was involved in the creation of the initial qEEG database (Monastra VJ et al, Appl Psychophysiol Biofeedback 2002;27(4):231–249).

While qEEG remains a controversy, this recent German study makes neurofeedback sound at least interesting for some patients. The practicalities are daunting, however. First, it is not generally covered by insurance, and the number of trainings is high—between 40 and 60 needed to obtain the results in the German study. If each session costs between $100 and $200 dollars, the total cost will be $4,000 to $12,000 per patient! Second, finding a reliable provider can be another hurdle. There is a certification process available, but not a uniform standard for certification or practice. (For more information on this, see the Biofeedback Certification Institute of America (BCIA) at www.bcia.org.)

Enter the toy industry. For about the same cost as a single session, interested patients can obtain a Star Wars Force Trainer (Uncle Milton), a toy that links a headset to a blower: the more you “concentrate” the harder the blower blows the ball in the air. The makers of the headset, NeuroSky, do not state what exactly is being measured by the headset. The website describes a “proprietary algorithm” that is able to detect “attention” and “meditation” states, and is able to measure alpha, beta, gamma, delta, and theta bandwidths. The toy makes no claims at all, unless you count the one about padawan training.

This same headset is used in the MindFlex game by Mattel, a similar toy that incorporates an obstacle course into the ball blower combo. It costs closer to a session and a half and makes no padawan training claims.

From there, a variety of neurofeedback equipment is available, generally tending toward the more expensive. Whether it works is open to speculation, especially as the specific markers used for feedback are not well described.