New Review of Neurofeedback Treatment for ADHD — Current State of the Science

By: Dr. David Rabiner

Neu­ro­feed­back — also known as EEG Biofeed­back — is an approach for treat­ing ADHD in which indi­vid­u­als are pro­vided real-time feed­back on their brain­wave activ­ity and taught to alter their typ­i­cal EEG pat­tern to one that is con­sis­tent with a focused and atten­tive state. Accord­ing to neu­ro­feed­back pro­po­nents, this often results in improved atten­tion and reduced hyperactive/impulsive behavior.
Sev­eral years ago I sum­ma­rized the sci­en­tific sup­port for neu­ro­feed­back treat­ment — see here - and noted that although pos­i­tive find­ings had been reported in mul­ti­ple pub­lished stud­ies, lim­i­ta­tions of these stud­ies led many researchers to regard neu­ro­feed­back as a promis­ing, but unproven treatment.
The Amer­i­can Psy­cho­log­i­cal Asso­ci­a­tion has estab­lished a 5-level sys­tem for grad­ing the evi­dence in sup­port of men­tal health treat­ments. These grades, and their asso­ci­ated lev­els of research sup­port, are as follows:
Level 1 Not Empir­i­cally Supported 
Sup­ported only through anec­do­tal evi­dence or non-peer reviewed case-studies.
Level 2 Pos­si­bly Efficacious 
Shown to have a sig­nif­i­cant impact in at least one study, but the study lacked a ran­dom­ized assign­ment between controls.
Level 3 Prob­a­bly Efficacious 
Shown to pro­duce pos­i­tive effects in more than one clin­i­cal, obser­va­tional wait list or within-subject or between-subject study.
Level 4 Efficacious 
Shown to be more effec­tive than a no-treatment or placebo con­trol group; the study must con­tain valid and clearly spec­i­fied out­come mea­sures, and it must be replic­a­ble by at least two inde­pen­dent researchers demon­strat­ing the same degree of efficacy.
Level 5 Effi­ca­cious and Specific 
Shown to be sta­tis­ti­cally supe­rior to cred­i­ble placebo ther­a­pies or to actual treat­ments, and it must be shown as such in two or more inde­pen­dent studies.
Using the grad­ing sys­tem above, and based on stud­ies pub­lished through 2005, the con­clu­sion reached by the Pro­fes­sional Advi­sory Board of CHADD was that evi­dence sup­port­ing neu­ro­feed­back treat­ment for ADHD war­ranted a Level 2, or ‘Pos­si­bly Effi­ca­cious’. You can read CHADD’s sum­mary state­ment at www.help4adhd.org/en/treatment/complementary/WWK6A and I believe you will find this to be of interest.
Based on a research base that includes more recently pub­lished stud­ies, how­ever, the con­clu­sions reached by Arn et. al., (2009) were far more pos­i­tive. These researchers con­ducted a meta-analysis of 15 stud­ies, 4 of which were reported to be ran­dom­ized con­trolled tri­als. Their con­clu­sion was 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.” This is a very dif­fer­ent con­clu­sion from the CHADD review and it is under­stand­able that par­ents, edu­ca­tors, and pro­fes­sion­als would be con­fused about the strength of the evi­dence base for neu­ro­feed­back treatment.
New Review Pro­vides Some Clarification
The Jour­nal of Atten­tion Dis­or­ders recently pub­lished an updated review of neu­ro­feed­back treat­ment for ADHD that helps clar­ify its sci­en­tific sup­port [Loft­house et. al., (2011). A review of neu­ro­feed­back treat­ment for ADHD. Jour­nal of Atten­tion Dis­or­ders, pub­lished online 16 Novem­ber 2011. DOI: 10.1177/1087054711427530]. The authors include sci­en­tists who have con­ducted research tri­als of neu­ro­feed­back and also been part of the Mul­ti­modal Treat­ment Study of ADHD (MTA Study), the largest ADHD treat­ment study ever con­ducted. They are thus well equipped, in my view, to pro­vide a thor­ough and objec­tive review of a com­pli­cated area.
The research base for their review was 14 stud­ies of neu­ro­feed­back treat­ment for chil­dren with ADHD in which par­tic­i­pants were ran­dom­ized to neu­ro­feed­back treat­ment or a con­trol con­di­tion. Eleven of these stud­ies were con­ducted between 2005 and 2010; this speaks to the strong accel­er­a­tion of neu­ro­feed­back research, which is a wel­come development.
Their review was lim­ited to those that ran­domly assigned chil­dren to treat­ment or con­trol con­di­tions which is an essen­tial ele­ment of rig­or­ous treat­ment stud­ies. For each study, the authors pro­vide a detailed cri­tique it’s strengths and lim­i­ta­tions. As a detailed review of the indi­vid­ual study cri­tiques is beyond what I can do here, below I sum­ma­rize the authors’ con­clu­sions on the state of the science.
Results Sum­mary
Treat­ment effects
When aver­aged across the stud­ies for which appro­pri­ate out­come data was avail­able, the over­all mean effect size (ES) was .79 for inat­ten­tion mea­sures, and .71 for hyperactivity/impulsivity mea­sures. These are in a range that would be con­sid­ered ‘large’ for inat­ten­tion and ‘mod­er­ate’ for hyperactivity/impulsivity and are below what is typ­i­cally reported for stim­u­lant med­ica­tion. Five of the stud­ies showed neu­ro­phys­i­o­log­i­cal changes that were spe­cific to neu­ro­feed­back treat­ment. Over­all, these results are con­sis­tent with ben­e­fi­cial effects of neu­ro­feed­back treat­ment for ADHD.
Sum­mary of study limitations
The authors iden­ti­fied 5 dif­fer­ent lim­i­ta­tions that under­mine the con­clu­sions about neu­ro­feed­back treat­ment effi­cacy that can be made.
1. Min­i­mal use of Triple Blinding 
The ideal study would be one where chil­dren, par­ents and/or teach­ers who rate children’s behav­ior before and after treat­ment, and clin­i­cians don’t know whether the child received active treat­ment. This elim­i­nates — or at least strongly reduces — the like­li­hood that appar­ent ben­e­fits asso­ci­ated with neu­ro­feed­back can be explained by expec­ta­tions that the child would benefit.
Only 4 of the 14 stud­ies uti­lized triple blind pro­ce­dures, how­ever, and in 6 of the stud­ies none of these 3 sources was blind.
2. Nature of Con­trol Group 
The strongest neu­ro­feed­back treat­ment study would be one that used ‘sham’ treat­ment for chil­dren ran­dom­ized to the con­trol group, i.e., par­tic­i­pants receive feed­back that is not linked to the EEG state that is the focus of actual train­ing. The ben­e­fit of this is that — in the­ory — it keeps chil­dren, par­ents, and clin­i­cians blind to whether real treat­ment is being pro­vided, thus elim­i­nat­ing poten­tial biases to the out­come rat­ings they provide.
For the 14 stud­ies review, how­ever, only 4 employed sham treat­ment. And, of those 4, only 1 used what was felt to be a truly cred­i­ble ‘sham’. In the absence of a cred­i­ble ‘sham’ treat­ment, con­duct­ing a ‘triple blind’ study is not possible.
The other stud­ies either used ‘wait list’ con­trols or com­pared neu­ro­feed­back treat­ment to a dif­fer­ent type of cog­ni­tive train­ing. The use of wait list and alter­na­tive treat­ment con­trol groups are preva­lent in the treat­ment lit­er­a­ture, but are less able than a true ‘sham’ con­di­tion to unequiv­o­cally estab­lish that treat­ment gains asso­ci­ated with neu­ro­feed­back are attrib­ut­able to the feed­back chil­dren receive on their EEG state.
3. Insuf­fi­cient iden­ti­fi­ca­tion, mea­sure­ment, and con­trol of con­comi­tant treatments
Chil­dren par­tic­i­pat­ing in these stud­ies were fre­quently receiv­ing other treat­ments as well, either med­ica­tion, psy­chother­apy, or edu­ca­tional inter­ven­tions. Because the pres­ence and changes in con­comi­tant treat­ments tended not to be care­fully mon­i­tored, how­ever, pos­i­tive change asso­ci­ated with neu­ro­feed­back may have been caused, or at least influ­enced in some way, by unre­ported changes in these other treatments.
4. Gen­eral lack of post-treatment follow-up 
Fol­low­ing chil­dren beyond the end of neu­ro­feed­back treat­ment is crit­i­cal for deter­min­ing long-term effi­cacy and/or the need for booster ses­sions. How­ever, only 3 of the stud­ies included a post-treatment fol­low up of neu­ro­feed­back. And, in these stud­ies, the pro­ce­dures for assess­ing the sus­tain­abil­ity of treat­ment ben­e­fits were judged to be com­pro­mised. Thus, the authors con­clude that the dura­tion of any gains asso­ci­ated with neu­ro­feed­back remains largely unknown.
5. Lim­ited atten­tion to pos­si­ble adverse side effects
Although neu­ro­feed­back is described as safe and with­out side effects, only 1 study actu­ally mon­i­tored and reported adverse events that chil­dren and par­ents related to treat­ment. Although no such effects were found, some have argued that all truly effec­tive treat­ments pro­duce some side effects in some per­cent­age of indi­vid­u­als who receive them. Thus, rather than not attend­ing to this pos­si­bil­ity in neu­ro­feed­back stud­ies because the treat­ment is assumed to be safe, the authors sug­gest that this is an area where greater scrutiny is warranted.
Over­all Summary
Based on their review of the lit­er­a­ture, the authors argue that “…due to the lack of blind­ing and sham con­trol con­di­tions in ran­dom­ized stud­ies” neu­ro­feed­back treat­ment for ADHD should not be con­sid­ered ‘Effi­ca­cious and Spe­cific’ as was con­cluded in the 2009 review by Arn and his colleagues.
Instead, they believe that a grade of 3 on the APA evi­dence scale, which cor­re­sponds to ‘Prob­a­bly Effi­ca­cious’ is war­ranted. They note that a large mul­ti­site triple-blind sham-controlled Ran­dom­ized Con­trolled Trial is needed to set­tle the issue.
Clearly, it is pos­si­ble to review the same evi­dence and reach a dif­fer­ent con­clu­sion. Some would argue that the authors are overly cau­tious in the evi­dence grade they assign and that more is being required of neu­ro­feed­back than of other ADHD treat­ments. For exam­ple, although the long-term ben­e­fits of neu­ro­feed­back treat­ment may remain rel­a­tively unknown, evi­dence on the long-terms ben­e­fits of med­ica­tion treat­ment is also limited.
One could also argue that requir­ing a triple-blind trial with a cred­i­ble sham con­di­tion is unrea­son­able because this is a higher stan­dard than that employed most psy­chother­apy out­come research. In stud­ies to estab­lish the effi­cacy of behav­ioral treat­ment for ADHD, for exam­ple, a triple blind trial is not pos­si­ble because clin­i­cians know what treat­ment they are pro­vid­ing and par­ents will know what treat­ment their child is receiv­ing. Despite this, behav­ior ther­apy is con­sid­ered a strong evidence-based treat­ment for ADHD.
In response to this objec­tion, the authors argue that the high­est stan­dard of sci­en­tific rigor should be required for any treat­ment offered to the pub­lic for which triple blind stud­ies are pos­si­ble (they are not pos­si­ble for behav­ior ther­apy), and which are not pre­cluded by strong eth­i­cal con­sid­er­a­tions. They note that this is espe­cially true for neu­ro­feed­back, as such a study is pos­si­ble and the treat­ment requires sub­stan­tial time, effort, and expense.
Some Final Thoughts
My view is con­sis­tent with the authors. I would very much like to see the type of study they call for and believe the evi­dence grade they sug­gest of ‘Prob­a­bly Effi­ca­cious’ is appro­pri­ate. Hav­ing this con­clu­sion pub­lished in a sci­en­tific jour­nal that does not focus on neu­ro­feed­back research rep­re­sents sig­nif­i­cant progress for the field as it was not too long ago that a com­monly held view seemed to be that there was lit­tle if any cred­i­ble evi­dence sup­port­ing this treatment.
It is also impor­tant to rec­og­nize that what remains unclear is not whether chil­dren with ADHD who receive care­fully admin­is­tered neu­ro­feed­back will gen­er­ally derive some ben­e­fit — the stud­ies reviewed in this arti­cle estab­lish that — but, rather, why does ben­e­fit occurs. Here is what the authors say:
“…due to the lack of con­trols, it is unclear as to whether the large ESs for impul­siv­ity and inat­ten­tion and the medium ES for hyper­ac­tiv­ity are due to the active com­po­nent of EF and/or non­spe­cific treat­ment factors.”
In other words, the research estab­lishes that neu­ro­feed­back treat­ment yields ben­e­fits for core ADHD symp­toms but is not clear on what explains those ben­e­fits. Is it the spe­cific feed­back on EEG activ­ity and learn­ing to con­trol that activ­ity that pro­duces the gains? Or do non­spe­cific fac­tors asso­ci­ated with the treat­ment, e.g., expectancy effects, clin­i­cian atten­tion, praise for the effort involved, etc., that actu­ally accounts for the gains?
This is the impor­tant sci­en­tific ques­tion that remains to be answered. In the mean­time, how­ever, the research reviewed here indi­cates that if par­ents obtain high qual­ity neu­ro­feed­back treat­ment for their child there is a rea­son­able basis for expect­ing that ben­e­fits will occur. The deci­sion to do so should be made with the knowl­edge that med­ica­tion treat­ment and behav­ioral ther­apy would be regarded as hav­ing stronger research sup­port at this time.
To dis­miss neu­ro­feed­back treat­ment sim­ply as ‘unproven’, how­ever, ignores the con­sid­er­able research on this approach that has been con­ducted. Help­ing fam­i­lies bet­ter under­stand the strengths and lim­i­ta­tions of this research can enable them to make a bet­ter informed deci­sion about whether to con­sider this treat­ment option for their child