Tuesday, June 6, 2017

Neurofeedback and ADHD

Neurofeedback, the use of feedback using features of EEG activity, to change the brain’s state, has been around since the 1970s. It holds unique appeals and challenges for clinicians, researchers, and the public and has been particularly popular as a complementary treatment for ADHD. In a typical neurofeedback session, you sit facing a computer or video screen with 1-4 EEG electrodes affixed to your head playing a video game with your mind. Depending on the software and goals, you might be instructed to levitate CGI rocks on a screen or keep a video-game character running around obstacles using your mind. The success depends on whether or not you reach or exceed a target level of a frequency range, or a particular ratio of frequencies in the EEG.

Does this kind of neurofeedback work? While some studies show that it does, in some sense, to some degree, recent meta-analyses and placebo-controlled studies strongly suggest that it “does not.” But, to regard the situation in terms of such a dichotomy may obscure what these studies imply not only about neurofeedback for ADHD, but also about the nature of ADHD, the relations between EEG and cognition, and the processes of conditioning and learning. The question of neurofeedback’s value is a significant one, and the seeming contradiction among studies raises implications and questions, possibly paradigm-shaking, for theory and practice in both neuroscience and complementary medicine…

The first work relevant to ADHD was the use of SMR conditioning as an anti-convulsant, first in cats, then humans (Tan et al. 2009 in Arns et al. 2014). SMR, the “sensory-motor-rhythm” (also called mu waves in humans) are spindles at 13-15 Hz (in the beta range) that occur in the sensorimotor cortex, most prominently in the absence of movement. (It was Barry Sterman who then famously discovered, by accident, that cats who had gone through SMR conditioning were less likely to have seizures when dosed with hydrazine, a component of rocket fuel (a NASA study); it was confirmed that SMR therapy could reduce seizures in human epilepsy, and that became one of the early successes of neurofeedback).  Studies in the 70s first showed that training children to control SMR decreased hyperactivity and distractability. The cat-SMR studies suggested that SMR training increased density of sleep-spindles and depth of sleep–hinting that, in some people, ADHD  may be related to sleep deprivation. SMR / beta-increase is still among the most popular and well supported treatment protocols for ADHD.

70s research also first showed that the apparent meanings of recognized EEG frequencies were insufficient for therapeutic purposes; for example, increasing alpha-waves, , supposedly the frequency of calm, had no benefit for hyperactivity.

Around the same time, a second ADHD-relevant signal was discovered—the Contingent Negative Variation or CNV.  CNV is not a matter of frequency, but rather a a kind of Slow Cortical Potential (SCP) a relatively slow (e.g. 300 millisecond) and subtle negative shift in the of the EEG from some cortical area which seems to show anticipation of an action or stimulus.  A negative correlation was discovered between the amount of CNV and the severity of ADHD symptoms and the potential to increase CNV or SCP through neurofeedback gained support. Subjects are trained to shift the SCP negatively (for attention) or positively (for relaxation), and this is a common protocol for ADHD therapy today.

Another group at UCLA (Abney, et al.) collected a database of EEG from ‘normal, healthy’ people and used early computers to average and compare to groups of people diagnosed with ADHD, thereby deriving the idea of the theta-beta ratio (TBR), the ratio between the fraction of EEG activity in the theta frequency to the fraction of activity in the beta range, as both a diagnostic and target for ADHD neurofeedback.  The TBR is the first FDA approved EEG based diagnostic test for ADHD and is also a popular EEG neurofeedback metric.

Conclusions regarding the efficacy of neurofeedback for ADHD are based most importantly on subjects’ behavior outside of treatment sessions, often as rated by teachers, and/or parents. Some studies have had blind raters, others not; in many studies the degree of blindness can only be inferred. Some research uses tools to test more specific dimensions of cognitive performance, but conclusions focus on the three definitive symptoms of ADHD–distractability, impulsivity, and hyperactivity—which are normally rated, rather than tested. And of course, the subjects’ performance during the neurofeedback process is a crucial dimension of the research but also does not substantiate efficacy as a medical treatment.

Studies in the 90s and 2000s, and a 2009 meta-analysis (Arns et al), seemed to confirm that both TBR and SCP therapy can bring significant improvements in attention and impulsivity, and more long-lasting effects than medication. Some studies seemed to show an advantage for SCP over TBR in relation to hyperactivity, but this still needs confirmation.

However in 2012-2013 placebo-controlled studies were published showing no significant difference between real and sham neurofeedback (where the subject’s EEG is not generating the feedback)(e.g. Arnold et al. 2012). Further meta-analyses in 2013 and 2014 looked at “blindness” in the studies and seemed to show that ADHD neurofeedback only has a significant effect only if non-blind parents and teachers, rather than blind raters, evaluate the behavioral improvements.

Yet clinicians often swear by it as do some patients.

Together, the research on ADHD neurofeedback seems to suggest that neurofeedback for ADHD can work, but likely not in the straightforward manner previously assumed; its effects may be to a large degree unrelated to the EEG aspect of the treatments! Or they may be pointing us towards new understandings of the relations between EEG and cognition. Here are some specific questions regarding . . .

ADHD: Are there sub-types of ADHD with different causes and neural profiles? Evidence hints that some may be related to sleep-derivation, others hyperarousal.

EEG: What causes differences in TBR and SMR aside from ADHD? Is a standard average TBR a good diagnostic, or does it mask the importance of individual variation? Evidence suggests that theta-rhythms might play a crucial role in executive control, so are TBR protocols, which involve theta-reduction, sabotaging their own goals, reducing the value of beta-increase for ADHD?

Neurofeedback: How important is learning theory to the efficacy of neurofeedback? How important are specific types of feedback? How much is our lack of standard treatment protocols clouding study results? How important is parenting style or other “non-specific” factors that might explain how neurofeedback can work other than through its presumed mechanism?

These questions have been the targets of relatively little or no research. But the fact that we are asking them shows that we have learned much and the next few years will surely uncover more puzzle-pieces with further glimpses of the emerging picture.

References at link.

1 comment:

  1. Cortese S, Ferrin M, Brandeis D, Holtmann M, Aggensteiner P, Daley D, Santosh P, Simonoff E, Stevenson J, Stringaris A, Sonuga-Barke EJ; European ADHD Guidelines Group (EAGG). Neurofeedback for Attention-Deficit/Hyperactivity Disorder: Meta-Analysis of Clinical and Neuropsychological Outcomes From Randomized Controlled Trials. J Am Acad Child Adolesc Psychiatry. 2016 Jun;55(6):444-55.

    We performed meta-analyses of randomized controlled trials to examine the effects of neurofeedback on attention-deficit/hyperactivity disorder (ADHD) symptoms and neuropsychological deficits in children and adolescents with ADHD.
    We searched PubMed, Ovid, Web of Science, ERIC, and CINAHAL through August 30, 2015. Random-effects models were employed. Studies were evaluated with the Cochrane Risk of Bias tool.
    We included 13 trials (520 participants with ADHD). Significant effects were found on ADHD symptoms rated by assessors most proximal to the treatment setting, that is, the least blinded outcome measure (standardized mean difference [SMD]: ADHD total symptoms = 0.35, 95% CI = 0.11-0.59; inattention = 0.36, 95% CI = 0.09-0.63; hyperactivity/impulsivity = 0.26, 95% CI = 0.08-0.43). Effects were not significant when probably blinded ratings were the outcome or in trials with active/sham controls. Results were similar when only frequency band training trials, the most common neurofeedback approach, were analyzed separately. Effects on laboratory measures of inhibition (SMD = 0.30, 95% CI = -0.10 to 0.70) and attention (SMD = 0.13, 95% CI = -0.09 to 0.36) were not significant. Only 4 studies directly assessed whether learning occurred after neurofeedback training. The risk of bias was unclear for many Cochrane Risk of Bias domains in most studies.
    Evidence from well-controlled trials with probably blinded outcomes currently fails to support neurofeedback as an effective treatment for ADHD. Future efforts should focus on implementing standard neurofeedback protocols, ensuring learning, and optimizing clinically relevant transfer.