How Strong is the Research Support for Neurofeedback in Attention Deficits?

(Edi­tor’s Note: Neu­ro­feed­back is one of the tech­nolo­gies that peo­ple often ask us about.  It is a promis­ing inter­ven­tion in a vari­ety of areas, and has got sig­nif­i­cant trac­tion in help­ing kids with ADD/ ADHD. Now, giv­en the sig­nif­i­cant cost it pos­es for par­ents, we need to ask the ques­tion: “How Strong is the Research Sup­port for Neu­ro­feed­back Treat­ment of Chil­dren with ADHD”? We are hon­ored to present the thoughts of Duke Uni­ver­si­ty’s Dr. David Rabin­er, a lead­ing author­i­ty on the field, on that impor­tant issue. As a bonus, you will enjoy his detailed descrip­tion and sug­ges­tions of how to design a high-qual­i­ty sci­en­tif­ic study.)

(Update as of March 2009: Dr. David Rabin­er has writ­ten an update to the arti­cle below based on a new­er study. You can read it click­ing on link: New Study Sup­ports Neu­ro­feed­back Treat­ment for ADHD)
————————————–

How Strong is the Research Sup­port for Neu­ro­feed­back Treat­ment of Chil­dren with ADHD?

Nei­ther of the two promi­nent approach­es to treat­ing ADHD — med­ica­tion treat­ment and behav­ior ther­a­py — are expect­ed to effect long term changes in the child. Med­ica­tion treat­ment induces short-term changes in brain activ­i­ty that is asso­ci­at­ed with a reduc­tion in symp­toms for many indi­vid­u­als. Behav­ior ther­a­py attempts to cre­ate a set of envi­ron­men­tal con­tin­gen­cies that pro­mote desired behav­ior in the child, but which is unlike­ly to endure when those con­tin­gen­cies are removed.

In recent years, researchers have begun devot­ing greater atten­tion to the pos­si­bil­i­ty that chil­dren — and adults — may be pro­vid­ed with par­tic­u­lar kinds of expe­ri­ences that may induce alter­ations in brain func­tion­ing that are asso­ci­at­ed with more endur­ing changes, i.e., they do not dis­si­pate as soon as treat­ment ends.

Neu­ro­feed­back — also known as EEG Biofeed­back — is reflec­tive of this approach and has a his­to­ry that goes back at least 30 years. It involves pro­vid­ing indi­vid­u­als with real-time feed­back on their brain­wave activ­i­ty in order that they may learn to alter their typ­i­cal EEG pat­tern to one that is con­sis­tent with a focused and atten­tive state. Accord­ing to neu­ro­feed­back pro­po­nents, when this occurs, improved atten­tion and reduced hyperactive/impulsive behav­ior will result. Fur­ther­more, this change is like­ly to endure because it reflects a basic alter­an­tion in an impor­tant aspect of brain functioning.

In this arti­cle, I’ll take a look at some of the con­tro­ver­sy sur­round­ing neu­ro­feed­back treat­ment and attempt to high­light some of the strengths and lim­i­ta­tions in the exist­ing pub­lished research.

- Con­tro­ver­sy Sur­round­ing Neu­ro­feed­back Research — 

Neu­ro­feed­back treat­ment for ADHD has been a source of sub­stan­tial con­tro­ver­sy in the field for many years and remains so today. Although there are a num­ber of pub­lished stud­ies in which pos­i­tive results have been report­ed, many promi­nent ADHD researchers feel that giv­en sig­nif­i­cant lim­i­ta­tions to the design and imple­men­ta­tion of these stud­ies, neu­ro­feed­back should be con­sid­ered a promis­ing, but unproven treatment.

This posi­tion is sum­ma­rized in the CHADD fact sheet on alter­na­tive and com­ple­men­tary inter­ven­tions, which includes the fol­low­ing state­ment about neurofeedback:

“It is impor­tant to empha­size, how­ev­er, that although sev­er­al stud­ies of neu­ro­feed­back have yield­ed promis­ing results, this treat­ment has not yet been test­ed in the rig­or­ous man­ner that is required to make a clear con­clu­sion about its effec­tive­ness for AD/HD. The afore­men­tioned stud­ies can not be con­sid­ered to have pro­duced per­sua­sive sci­en­tif­ic evi­dence con­cern­ing the effec­tive­ness of EEG biofeed­back for ADHD.

Con­trolled ran­dom­ized tri­als are required before con­clu­sions can be reached. Until then, buy­ers should beware of the lim­i­ta­tions in the pub­lished sci­ence. Par­ents are advised to pro­ceed cau­tious­ly as it can be expen­sive — a typ­i­cal course of neu­ro­feed­back treat­ment may require 40 or more ses­sions — and because oth­er AD/HD treat­ments (i.e., mul­ti-modal treat­ment) cur­rent­ly enjoy sub­stan­tial­ly greater research support.”

You can find the com­plete CHADD fact sheet Here.

What Does an ‘Ide­al’ Treat­ment Study Look Like? 

Before review­ing some recent neu­ro­feed­back stud­ies, it would be use­ful to con­sid­er what an ‘ide­al’ treat­ment study entails. This will pro­vide a con­text against which recent neu­ro­feed­back stud­ies can be evaluated.

Ide­al­ly, treat­ment stud­ies are designed so that if pos­i­tive results are obtained, all pos­si­ble expla­na­tions for those results besides the treat­ment itself have been elim­i­nat­ed. This requires 2 basic ele­ments: ran­dom assign­ment and an appro­pri­ate con­trol group.

Ran­dom Assignment

Imag­ine that you are test­ing a new med­ica­tion treat­ment for ADHD with 50 chil­dren who have been care­ful­ly diag­nosed. In a ran­dom assign­ment study, whether each child is assigned to the treat­ment or con­trol con­di­tion is deter­mined by chance — you could flip a coin and give the med­i­cine to the ‘heads’ and noth­ing to the ‘tails’. This insures that any dif­fer­ences that might exist between chil­dren who get the med­ica­tion and those who don’t are pure­ly chance dif­fer­ences. At the end of the study, if those who received the med­i­cine are doing bet­ter, you could feel con­fi­dent that this is prob­a­bly due to the med­i­cine itself, and not to dif­fer­ences that may have been there before the treat­ment even started.

What if you did­n’t use ran­dom assign­ment, but let each child’s par­ents choose whether their child is in the treat­ment or con­trol group? In this case, it is pos­si­ble that chil­dren in the 2 groups dif­fered in impor­tant ways before the treat­ment began. If chil­dren who received the med­ica­tion were doing bet­ter at the end of the study, it might be because of dif­fer­ences that were there to start with.

For exam­ple, par­ents who chose the med­i­cine might be more will­ing to pur­sue oth­er ways to help their child than those who did­n’t. The fact that chil­dren who received the med­ica­tion were doing bet­ter at the end of the study might thus have noth­ing to do with the med­i­cine itself, but reflect oth­er things their par­ents were doing to help them. No mat­ter how hard you might try to rule out these oth­er pos­si­ble expla­na­tions — and I’m sure you can think of many oth­ers — you could nev­er do this with cer­tain­ty. Thus, I might rea­son­ably doubt that your new med­ica­tion is real­ly effective.

What about con­trol groups?

Even with the ran­dom assign­ment exam­ple above, how­ev­er, your study would still have an impor­tant prob­lem. Because chil­dren in the con­trol group received noth­ing, every­one knows who is being treat­ed and who isn’t. If you ask par­ents how their child is doing 4 weeks lat­er, this could very pos­si­bly bias their reports. Par­ents whose child received med­ica­tion may report their child is doing bet­ter sim­ply because they expect the med­i­cine to help. Par­ents of chil­dren in the con­trol con­di­tion may be biased against see­ing improve­ment because they know their child was not treated.

Thus, if par­ents of treat­ed chil­dren report­ed more improve­ment than par­ents of con­trol chil­dren, I could still rea­son­able ques­tion that the new med­ica­tion was tru­ly help­ful. Even if reports from teach­ers yield­ed sim­i­lar find­ings, I would argue that teach­ers might have learned when chil­dren were get­ting med­ica­tion. While this may be a low prob­a­bil­i­ty event, you could­n’t com­plete­ly rule it out. If I was a real skep­tic about your new med­ica­tion, your study would not be all that convincing.

The way around this is to cre­ate a sit­u­a­tion where no one knows — not the par­ents, child, teach­ers, researchers, or any­one else — whether the child is receiv­ing med­ica­tion. With med­ica­tion stud­ies, this is rel­a­tive­ly sim­ple to do: chil­dren in the con­trol con­di­tion are giv­en a place­bo pill that looks just like the real med­ica­tion but that has no active ingre­di­ents. Because no one knows who is get­ting med­ica­tion and who isn’t, pos­si­ble bias­es in par­ents’ and teach­ers’ reports at the end of the study are thus elim­i­nat­ed. This way, if chil­dren in the med­ica­tion con­di­tion are doing bet­ter at the end of the study than chil­dren get­ting place­bo, it is hard­er to doubt the ben­e­fits of your med­ica­tion. In fact, there would be no rea­son­able basis for such doubt, espe­cial­ly if you repeat­ed the study and found the same results.

Three Rep­re­sen­ta­tive Neu­ro­feed­back Studies

With this as a back­ground, let’s take a brief look at 3 recent neu­ro­feed­back stud­ies and see how they fare on the crit­i­cal dimen­sions of incor­po­rat­ing ran­dom assign­ment and an ‘ide­al’ con­trol group.

- Study 1 -

In an inter­est­ing study by Vince Monas­tra and his col­leagues [Monas­tra et al. (2002). The Effects of Stim­u­lant Ther­a­py, EEG Biofeed­back and Par­ent­ing Style on the pri­ma­ry symp­toms of ADHD. Applied Psy­chophys­i­ol­o­gy and Biofeed­back, 27, 249.] one hun­dred 6–19 year-olds with ADHD were treat­ed over 12 months. Some par­ents chose a treat­ment plan that includ­ed med­ica­tion, behav­ior ther­a­py, and school con­sul­ta­tion. Oth­er par­ents chose to add neu­ro­feed­back to their child’s treatment.

After 12 months, chil­dren whose treat­ment includ­ed neu­ro­feed­back back were doing bet­ter than the oth­er chil­dren accord­ing to both par­ents and teach­ers. They also showed ‘nor­mal’ EEG scans while the oth­er chil­dren still had the EEG pat­tern char­ac­ter­is­tic of ADHD. Even more impres­sive was that neu­ro­feed­back treat­ed chil­dren main­tained these ben­e­fits after med­ica­tion was dis­con­tin­ued for a week. You can read a detailed review of this study Here.

- Study 2 -

In a study by Fuchs et al (2003), par­ents of thir­ty-four 8–12-year-old chil­dren with AD/HD chose either stim­u­lant med­ica­tion or neu­ro­feed­back treat­ment for their child. After 3 months, chil­dren in both groups showed sig­nif­i­cant and com­pa­ra­ble reduc­tions in AD/HD symp­toms accord­ing to par­ents and teach­ers. Lab­o­ra­to­ry tests of atten­tion also showed equiv­a­lent improve­ment. A com­pre­hen­sive review of this study is avail­able at Here

- Study 3 -

In an espe­cial­ly inter­est­ing study, (Levesque, J., Beau­re­gard, M., & Men­sour, B. 2006. Effect of neu­ro­feed­back train­ing on the neur­al sub­strates of selec­tive atten­tion in chil­dren with AD/HD: A func­tion­al mag­net­ic res­o­nance imag­ing study. Neu­ro­science Let­ters, 394, 216–221.) twen­ty 8–12-year-old chil­dren with ADHD were ran­dom­ly assigned to receive neu­ro­feed­back treat­ment — 40 1‑hours ses­sions — or a wait-list con­trol condition.

At the end of the study, treat­ed chil­dren were doing sig­nif­i­cant­ly bet­ter than con­trol chil­dren accord­ing to par­ents. They also did bet­ter on sev­er­al objec­tive, lab­o­ra­to­ry mea­sures of atten­tion. Espe­cial­ly note­wor­thy was that fMRI scans used to mea­sure brain activ­i­ty dur­ing a com­plex cog­ni­tive task showed sig­nif­i­cant change for treat­ed chil­dren but no change for con­trol children.

You can find a detailed review of this study Here.

So, what’s wrong with these studies?

I sus­pect the answer to this ques­tion is obvi­ous. While the first 2 stud­ies yield­ed impres­sive results, nei­ther includ­ed ran­dom assign­ment. Thus, as dis­cussed above, it is not pos­si­ble to con­clude that it is the neu­ro­feed­back treat­ment specif­i­cal­ly that made the dif­fer­ence. Instead, dif­fer­ences that may have been present before treat­ment began and/or after treat­ment start­ed could be respon­si­ble. While this may be unlike­ly, it can’t be entire­ly discounted.

The third study includ­ed ran­dom assign­ment so this is not a prob­lem. It should be not­ed, how­ev­er, that with only twen­ty par­tic­i­pants, treat­ment and con­trol groups are more like­ly to dif­fer at the start of the study than if a larg­er sam­ple had been ran­dom­ly assigned.

The big­ger prob­lem is that the con­trol con­di­tion is not a very strong one in that every­one knew who received neu­ro­feed­back treat­ment and who did not. This may have biased par­ents’ rat­ings, although it is dif­fi­cult to argue that it could have biased chil­dren’s per­for­mance on the lab-based atten­tion tests or on the fMRI scan. Per­haps, how­ev­er, it was­n’t the spe­cif­ic feed­back on EEG states that neu­ro­feed­back pro­vid­ed, but sim­ply the atten­tion chil­dren received dur­ing the 40 hours of train­ing, that was the impor­tant fac­tor. Although unlike­ly in my opin­ion, this can’t be def­i­nite­ly ruled out.

What would an ‘ide­al’ neu­ro­feed­back study look like?

This may be pret­ty clear now as well. The most con­clu­sive test of neu­ro­feed­back treat­ment would include ran­dom assign­ment and a con­trol con­di­tion that close­ly matched the neu­ro­feed­back con­di­tion. For exam­ple, chil­dren could receive video game coach­ing from a sup­port­ive adult for the same time peri­od. Or, even bet­ter, they could do exact­ly what chil­dren get­ting the neu­ro­feed­back were doing but not receive direct feed­back on their EEG states. If pos­i­tive treat­ment results were still found, it would indi­cate that obtain­ing EEG feed­back and learn­ing to con­trol one’s EEG state is why neu­ro­feed­back treat­ment works.

I’ve been told by peo­ple who know much more about neu­ro­feed­back than I that such a study, while dif­fi­cult to do, is tech­ni­cal­ly pos­si­ble. I am not aware of such a study hav­ing been pub­lished; if you are, please let me know.

I should point out that there would be impor­tant eth­i­cal con­cerns with such a study. Neu­ro­feed­back treat­ment typ­i­cal­ly occurs over a peri­od of months. Chil­dren who were receiv­ing ‘sham’ neu­ro­feed­back would be get­ting a treat­ment that no one expect­ed to be help­ful for a sus­tained time period.

A more rea­son­able alter­na­tive might thus be to pro­vide chil­dren with a treat­ment of known effi­ca­cy — such as med­ica­tion — dur­ing the neu­ro­feed­back tri­al. If this were giv­en to chil­dren in the real and sham con­di­tions it would not cre­ate a con­found because both groups would be receiv­ing it. After neu­ro­feed­back was com­plet­ed, one could see whether chil­dren who got the real treat­ment were doing bet­ter — i.e., did neu­ro­feed­back add any­thing to med­ica­tion — and whether these ben­e­fits per­sist­ed after med­ica­tion was tem­porar­i­ly stopped.

What can we con­clude about neu­ro­feed­back until such a study is done?

It would be ter­rif­ic if a study like the one out­lined above were com­plet­ed. Unfor­tu­nate­ly, how­ev­er, I think it is a real pos­si­bil­i­ty, how­ev­er, that we will nev­er see such a study. This would be a large and expen­sive under­tak­ing and obtain­ing fund­ing for it would not be easy.

In the inter­im, my own view is that exist­ing sup­port for neu­ro­feed­back should not be so read­i­ly dis­count­ed because of the study lim­i­ta­tions dis­cussed above. Here’s why I think this is the case.

The way treat­ment was ‘assigned’ in the first 2 stud­ies sum­ma­rized is very sim­i­lar to what hap­pens when par­ents seek treat­ment for their child. That is, par­ents inves­ti­gate dif­fer­ent options and decide which one they want for their child.

This is what hap­pened in these stud­ies — some par­ents chose neu­ro­feed­back for their child and some did not. When neu­ro­feed­back was cho­sen, chil­dren were found to ben­e­fit on both ‘sub­jec­tive’ par­ent reports as well as on more ‘objec­tive’ assess­ments. This does not mean that neu­ro­feed­back would ‘work’ for chil­dren ran­dom­ly assigned to receive it. It also does not mean that rea­son neu­ro­feed­back works by pro­vid­ing spe­cif­ic train­ing in learn­ing how to man­age one’s EEG state. It is for these rea­sons that neu­ro­feed­back is under­stand­ably regard­ed as an unproven treat­ment approach for ADHD at this time by many ADHD researchers.

How­ev­er, these stud­ies do pro­vide a sol­id basis for sug­gest­ing that if par­ents choose to pur­sue neu­ro­feed­back for their child, there is a rea­son­able chance that their child will ben­e­fit even though we can’t be sure that it is the spe­cif­ic EEG train­ing that is respon­si­ble for the ben­e­fits. Thus, although the effi­ca­cy of neu­ro­feed­back has yet to be con­clu­sive­ly con­firmed in a ran­dom­ized, place­bo-con­trolled tri­al, it is impor­tant to place this lim­i­ta­tion in the con­text of the sup­port­ive research evi­dence that has been accumulated.

Pro­vid­ing this con­text can help fam­i­lies bet­ter under­stand the strengths and lim­i­ta­tions of the exist­ing research on neu­ro­feed­back and enable them to make a bet­ter informed deci­sion about whether to con­sid­er this treat­ment option for their child.

– Dr. David Rabin­er is a child clin­i­cal psy­chol­o­gist and Direc­tor of Under­grad­u­ate Stud­ies in the Depart­ment of Psy­chol­ogy and Neu­ro­science at Duke Uni­ver­sity. He pub­lishes Atten­tion Research Update, an online newslet­ter that helps par­ents, pro­fes­sion­als, and edu­ca­tors keep up with the lat­est research on ADHD, and teach­es the online course  How to Nav­i­gate Con­ven­tion­al and Com­ple­men­tary ADHD Treat­ments for Healthy Brain Devel­op­ment.

Oth­er arti­cles by Dr. Rabiner:

Promis­ing Cog­ni­tive Train­ing Stud­ies for ADHD.

Mind­ful­ness Med­i­ta­tion for Adults & Teens with ADHD.

Work­ing Mem­o­ry Train­ing for Adults.

Self-Reg­u­la­tion and Barkley’s The­o­ry of ADHD.