New Efficacy Document

by Siegfried Othmer | June 20th, 2008

The AAPB has just published the 2008 edition of its publication “Evidence-Based Practice in Biofeedback and Neurofeedback“, written by Carolyn Yucha and Doil Montgomery. This is an update on the 2004 edition. The new publication spruces up the reference list and adds new material. It now comes to 81 pages, whereas the first one was only 48. Yet there is a lot of commonality between the two versions, with most of the text just carrying over. There have been some steps forward, some steps backward, and some glaring problems persist.

The first problem I have is with the Alcoholism/Substance Abuse section. Even by their own cramped criteria, the combination of biofeedback and neurofeedback (that is to say, feedback-based self-regulation strategies) meet criteria for ‘Efficacious.’ There is first of all the first Peniston study, and then there is the CRI-Help study. That meets the formal criterion of two independent controlled studies. Multi-year follow-up was done in both of those studies that confirmed the results. These findings were essentially corroborated by the Burkett effectiveness study, in which substantial recovery was documented in the most difficult population imaginable–homeless crack addicts–using alpha/theta training. There is now no doubt whatsoever–none–that a neurofeedback strategy containing Alpha/Theta training contributes to better outcomes in addictions treatment over currently standard treatment.

In fact, if one looks at the CRI-Help outcome data with reference to the initial intake into the study, where the experimentals and the controls were matched for addiction severity, the controls (who got the conventional treatment) exhibited a failure rate of three out of four at the one-year point post-treatment. The experimentals, on the other hand (who also got neurofeedback) exhibited a success rate of three out of four. This kind of success on a large scale is essentially unprecedented in the field of addictions treatment. There was additional corroborative data in the form of TOVA scores, MMPI scores, and neurocognitive function. What doubt remains that needs to be resolved?

In asthma we have a condition for which no good, categorical remedy is available within standard medicine. Medical remedies merely help to manage the condition. That being the case, different efficacy criteria should apply than in the case of ADHD, where stimulant medication is in place as the presumptive remedy. That is to say, if a bad situation can be made somewhat better, that should be sufficient to claim efficacy for EEG biofeedback / neurofeedback. The answer to the question of whether an asthma sufferer should avail himself of the self-regulation technologies is a resounding yes. We are routinely seeing the use of inhalers drop away as children come to train for various conditions.

The ADHD section listed only some 16 studies in 2004, and that list has expanded to 43 citations in 2008. More European work is now included, among other things. ADHD is one of the few categories for which efficacy is acknowledged in this document.

Autism is listed as a condition for which biofeedback / neurofeedback is possibly efficacious, which again misrepresents the evidence. There is no listing of the early study by Betty Jarusiewicz, for example. In 2004 Autism was listed as ‘Not sufficiently investigated,’ and that designation was unfortunately abandoned in the new edition. We already know from clinical data that autism is highly responsive to neurofeedback. It’s just that formal studies are still in the pipeline right now.

Autoimmune disorders are not listed as a category at all. Perhaps that’s just as well, since these would have gotten another low rating on the efficacy scale. As with asthma, resort to a feedback-based self-regulation strategy is strongly to be recommeded for all such conditions. Sometimes the difference it makes is marginal; sometimes it is dramatic. One cannot yet predict outcomes in advance; the techniques simply have to be tried. The benefits are nearly always worthwhile.

Cerebral palsy is another condition where a systematic ‘cure’ is not really in prospect. But as with asthma and autoimmune conditions, brain-training should be resorted to as a matter of course. There is simply no question that cerebral palsy symptoms can be improved with neurofeedback; it is merely a question of how far one can go with any specific case. There is also the fact that cerebral palsy sufferers are subject to all the other conditions of disregulation just like everyone else. So even if the explicit cerebral palsy symptoms are recalcitrant, one may still be able to improve quality of life significantly in other ways.

It is gratifying to see that Chronic Pain has been moved up from Probably Efficacious to Efficacious. The number of citations has also increased from 5 to 27, so more serious attention was given to this condition. Here we have possibly the most obvious case in point of a condition where a categorical remedy is not available, but where symptom severity is clearly amenable to a feedback-based self-regulation strategy. And given the level of misery in which these people find themselves, all possible help should be provided.

The category of Eating Disorders has been relabelled from ‘Not Sufficiently Investigated’ to ‘Not Empirically Supported.’ Empirical evidence is, of course, quite otherwise; it’s just not contained in formal, published studies. And as for formal studies, Marvin Sams and Peter Smith reported on their results of several years of work with neurofeedback at an eating disorders treatment center, and the results were quite unambiguously favorable to neurofeedback. Unfortunately the paper never saw print. Intractable eating disorders are among those conditions that are intimately tied to an early childhood history of psychological trauma, and all such conditions should be treated with feedback as part of the therapy.

Epilepsy was listed under ‘Probably Efficacious’ in the 2004 report, and it is finally listed properly under ‘Efficacious’ in the present document. How could such mislabeling possibly have happened in 2004, with a history of 18 formal studies covering a period of 25 years? The only new study added to the citations in 2008 was one using electrodermal feedback. Hmmmm. So biofeedback makes things real for these people?

Headache is listed under ‘Efficacious’ for adults and under ‘Probably Efficacious’ for children. Clinically we don’t see much of a difference in outcome between children and adults, and headaches are among the most responsive of symptoms when it comes to neurofeedback. It’s nice to see, though, that all types of headaches are being lumped together here under one heading. We really don’t see much difference in outcome between the different types of headaches. The number of citations in this section went from 10 in 2004 to 20 in 2008.

Insomnia is placed in the ‘Probably Efficacious’ bin. If one were to inquire about general understandings in the medical world about biofeedback, one would find that biofeedback is seen as applicable to anxiety syndromes, pain conditions, and sleep disorders. These are also the conditions for which insurance reimbursement is most likely to be approved. This is not the place where the biofeedback community should sound an uncertain trumpet.

The fact is that the whole litany of sleep disorders is responsive to neurofeedback, including not only insomnia (sleep onset difficulties and frequent waking), but also bed-wetting in children, nightmares, night terrors, nocturnal encopresis, restless leg syndrome, nocturnal myoclonus, night sweats, nocturnal flashbacks in PTSD, and of course nocturnal seizures. I’ve probably forgotten a few. These collectively tell a story of sleep manifesting disorders of disregulation more readily, likely because conscious means of state management are not available and because cortex is less tightly regulated in some sleep states. (There is one sleep disorder we haven’t affected much, and it is the circadian rhythm disorder where the person’s biological day slips or advances a given amount each day.)

PTSD is listed under the heading of ‘Possibly Efficacious,’ when in fact Peniston demonstrated efficacy for his combined biofeedback/neurofeedback approach all by himself in two controlled studies. The EMDR work with PTSD is essentially supportive of the neurofeedback claim as well, so between them we have the well-validated proposition that reinforcement of low EEG frequencies (whether by stimulation or reinforcement) can contribute to the resolution of trauma formations.

Tinnitus is listed under the heading of ‘Possibly Efficacious.’ Here we have yet another condition that is intimately tied to a real organic deficit (so that a ‘cure’ is not in prospect), but one where symptom severity is clearly influenced by the ‘state’ of the person, which can in turn be influenced with feedback. So feedback should be tried, pure and simple. No one can yet predict the outcome. Tinnitus can be thought about much like chronic pain. Amelioration of suffering is in prospect even if the tinnitus itself remains.

Finally, we come to the most egregious travesties in this whole compilation: Traumatic Brain Injury and Spinal Cord Injury. Neurofeedback efficacy for traumatic brain injury has been well established through clinical work now for some decades. It was the life’s work of Margaret Ayers, who despite her strange ways had a significant influence on this field. After decades of her epochal work with intractable traumatic brain injury, and now as an epitaph upon her early death, the field renders the official judgment that ‘probably’ it was not all a waste of time.

And it gets worse. A parallel situation exists in the matter of spinal cord injury. This was the life’s work of Bernard Brucker, following upon the pioneering work of John Basmajian in electromyography. No one who has ever sat through Bernie’s slide show can have any doubt whatsoever as to the effectiveness of his work. But now the field renders its official judgment that his life’s work is ‘Not Empirically Supported.’ Another disgraceful epitaph for a pioneer whom we lost this year.

Here we have the situation where the outcome of an application of the rules leads to false conclusions. The conclusions are most obviously false with regard to traumatic brain injury and spinal cord injury, but with a whole lot of other conditions besides. Now any scientist who finds the outcome of his hypothesis to be unsupported by the facts has the obligation to revisit his assumptions. In this case, lifetimes’ worth of clinical data must be considered the definitive ground truth.

The efficacy criteria that have been applied yield the wrong conclusions precisely where by some other reasonable criteria the evidence would appear to be strongest. These conditions have something in common. They are not strictly disorders of disregulation but they have organic deficits to contend with. So the population is highly heterogeneous and hence relatively unsuitable for group studies. Biofeedback and neurofeedback for these conditions must be classified as a rehabilitative technique. Rehabilitation techniques typically exhibit a learning curve, and if that is the case, one does not need a controlled study to document them. The learning curve is its own proof.

Implicit in the efficacy criteria is the assumption that a valid control group must be run in parallel with the experimentals. But this is an absurd constraint that is entirely without provable scientific merit in the case of traumatic brain injury and spinal cord injury. In actual fact, work with these conditions is always effectively a controlled design because the trainees come in with a long baseline history of disability. Therein lies the control, and that is all the control one needs for the rehabilitative technologies. I order to discount the work of Ayers and Brucker, somebody has to argue that sequential controls lead to unscientific conclusions, and that has not been done.

Finally, there are the errors of omission in this document. Because of the focus on clinical conditions, whole categories of neurofeedback work are left out. Here’s why they shouldn’t be: The efficacy criteria have as their basis the need to dispel the placebo hypothesis. Hence the need for controlled designs that take placebo effects into account. The placebo hypothesis has the fatal flaw of needing to be universal. The neurofeedback deniers are totalitarians in the sense that they need to deny neurofeedback validity across the board. As soon as that hypothesis is punctured at any point at all, the discussion can be shifted to the more useful question of what can actually be accomplished with neurofeedback for different conditions.

By buying into the efficacy criteria, the authors have burdened themselves with needing to confront the placebo hypothesis independently for all conditions. How unfortunate. There are two kinds of studies that effectively dispel the placebo hypothesis for neurofeedback. One is the work demonstrating major improvements in IQ, and the other is the work of Kirt Thornton with specific learning disabilities. One could even cite all of the improvements in CPT scores that neurofeedback practitioners have published as a third element of the proof.

Here we have a non-pathological category, namely the IQ, which can serve as a change measure without entanglement in the various issues surrounding the standard diagnostic categories. In particular, the discussion can be distanced from any particular claims, diagnostic thresholds, diagnostic complexity, etc. For these purposes, IQ is simply defined as what the IQ test measures.
The very first such study already proved the point. Michael Tansey found average IQ score improvements of about 20 points in mildly neurologically impaired kids. This report was followed then by our replication of Tansey, in the 1991 timeframe, where we
obtained an average of 23-point IQ score gains in 15 children and adolescents. This was a soundly done study, with lots of complementary measures (and independent testing).
There is no placebo model that explains major increases in IQ, and in Kirt Thornton’s work one can find all the specificity that anyone could want–targeted training that has the desired outcome with identified cognitive deficits. Finally, there is no placebo model that can explain systematic normalization of TOVA scores. So the same approach as with IQ can be taken with TOVA or QIKtest data. One can simply compile, in a meta-analysis sense, all of the studies that tracked the TOVA. This avoids any argument about whether the TOVA scores actually index ADHD, or even impulsivity and inattention. The TOVA is simply treated for these purposes as a stand-alone test of reaction time, reaction time variability, omission errors, and commission errors.  Importantly, it is a test that has no practice effects, and it is a computer-scored test where the tester’s judgment is not relevant.

Both in the case of the neurofeedback deniers and the efficacy criteria fetishists, we are contending with scientific fundamentalists that are fully as rigid in their obtuseness as any religious fundamentalist. Argument in these cases is somewhat pointless because these people don’t feel themselves at liberty to change their views. The efficacy criteria are by now in the catechism–it no longer matters what anybody thinks. We’re up against scientific robotic dolls. Every time you push their belly button, they squawk ‘placebo,’ or maybe even ‘blinded controlled design with random assignment.’

Nevertheless, let me take one more stab at the placebo doctrine because of its over-weening influence on our experimental designs. Even if the IQ data, etc., establish NF as a technique that is not reducible to a placebo effect, the placebo is certainly still in play in the individual case, so we have not really disposed of this multi-headed hydra.

Let us confine this discussion to what may be called the ‘real’ placebo effect. We here distinguish the real placebo effect from the usual expectancy factors, from ambiguity in assessment measures, from misattribution (attributing all of NF to the placebo, for example), and from the ‘natural course of the disease,’ the ebb and flow in symptom severity that characterizes many conditions of interest. What is left, then, for the real placebo effect? It is the autonomous change in self-regulation status! The real placebo effect is nothing more than self-regulation at work in the state of nature. It is always at work, and its efficacy can be modulated by psychological factors.

Now once we have found a way to affect the instrumentalities of self-regulation, why would one care if nature can sometimes accomplish the same thing on its own? That simply supports our argument; it does not diminish it. The placebo is such a potent concept because self-regulation is such a robust and ever-present mechanism. This only demonstrates the broad reach of brain plasticity, and that supports our case for doing biofeedback. What nature may accomplish randomly, we are now able to do systematically. And the proof of being able to work systematically can be adduced in an entirely self-referential way, by demonstrating learning curves and by demonstrating predictability and reproducibility of results.

The reliance on self-referential data is actually quite common within the field, as illustrated by the fact that negative effects attributable to feedback are almost always taken at face value. So, negative effects are always real, and positive effects always remain ephemeral. (In fact, if negative effects are accepted as real, it must be allowed that they also kill the placebo hypothesis.)

Here in Los Angeles the controversy about cloud-seeding has recently been revived. It was routinely done until three decades ago, when a major flood ensued that was blamed on the technique. Clearly cloud-seeding can have desirable effects under some conditions, and that is good enough. It does not even have to be better than placebo, i.e. the raw state of nature. It just has to be incrementally helpful in order to be worthwhile.

If we as a society decide not to use rain-modification technology because it is not clearly ‘better than placebo,’ we are no further along than the Indian custom of doing the rain dance. And every time the decision is made not to use neurofeedback because of this wretchedly misguided efficacy document, we are saying effectively, ‘better to let nature take its course and hope that self-regulation sets in spontaneously.’ We will, of course not hesitate to offer the proverbial sugar pill and all the moral support we can muster. Rain dance, anyone?

Siegfried Othmer, Ph.D.

Share your thoughts in the comments section below.

2 Responses to “New Efficacy Document”

  1. quovadis says:

    Interesting post. I tried to access the original standards, but the link to the AAPB web site is not working. It’s just says page not found. In fact the guidelines seem to have disappeared entirely from the AAPB website. I emailed them about this 4 days ago, and have not even received an acknowledgement. There’s no feedback form to avoid the spam blockage problem.
    So no wonder neurofeedback is not used as much as it should be, if the organisation responsible for promotion does not maintain an up to date website or respond to queries.

  2. Thanks for writing. I suspect that the old document was pulled from the website now that the new publication is available.

    You have certainly put your finger on a problem. The organization really does not see its role as the promotion of the field. There is a long tradition in the health care field that promotion is destructive of professional standards and values. That’s why it was so crucial for drug companies to obtain the right to market directly to the public. That has made health care marketing commonplace in our experience, but the professional communities remain quite ambivalent about going there.

    Siegfried Othmer

Leave a Reply