This policy is based in part on a 1997 Blue Cross Blue Shield Association Technology Evaluation Center (TEC) Assessment which concluded that there were inadequate data to permit conclusions regarding the health outcome effects of neurofeedback for any indication. (1) Among the 19 studies reviewed in the TEC Assessment, few were randomized controlled trials (RCTs), and those that were did not support the efficacy of neurofeedback in improving health outcomes. In addition, even among the RCTs, only 2 studies used appropriate control conditions. A literature review for the period between 1998 and October 2004 identified few relevant articles. (2-6) Periodic literature review updates using the MEDLINE database, performed most recently for the period of July 2010 through May 2011, indicate increasing interest in neurofeedback for a variety of conditions, although the scientific literature appears to be the most advanced for attention deficit/hyperactivity disorder (ADHD). (7,8) Relevant systematic reviews and key randomized or controlled trials of neurofeedback are described here.
Attention Deficit Hyperactivity Disorder
A 2005 review/meta-analysis used criteria from the Association for Applied Psychophysiology and Biofeedback (AAPB) and the International Society for Neuronal Regulation (ISNR) to assess the clinical efficacy of neurofeedback for attention deficit/hyperactivity disorder (ADHD). (9) The authors concluded that neurofeedback for ADHD was ranked at level 3 or "probably efficacious" on a scale of 1 to 5 (1 being not empirically supported and 5 being efficacious and specific). The authors noted that benefits were reported in the 5 randomized group studies (totaling 214 patients) included in their analysis; however, the ranking for neurofeedback for ADHD was based on the need for further studies controlled for patient and therapist factors that could unduly influence outcomes.
In 2009, Arns and colleagues published a meta-analysis on neurofeedback and ADHD, concluding that neurofeedback could be considered “efficacious and specific” for ADHD based on level 5 evidence. (8) Fifteen studies met criteria (either between-subject or within-subject design) and were included in the analysis. Initial analysis indicated heterogeneity in study results, which typically would preclude meta-analysis. For this paper, studies were removed from the analysis until heterogeneity was achieved. The adjusted analysis indicated similar effect sizes between neurofeedback and stimulant medication; however, this result was based on nonrandomized studies in which patients chose their treatment; this study design has a high potential for selection bias. (10) Four RCTs that utilized either a wait-list control or active control group were included in the meta-analysis. One of the studies is a German language report and another is an unpublished PhD thesis (total of 69 children); these have not been reviewed in this policy. The other 2 RCTs included in the systematic review are described below, including 20 and 94 children with ADHD, respectively. (11,12) Overall, the literature included in this meta-analysis is characterized by small, poor quality studies with high potential for bias. The findings of the meta-analysis are also limited by significant heterogeneity in study results and exclusion of studies due to heterogeneity. Details of the English language RCTs are described below.
A randomized study published in 2006 examined brain activity following neurofeedback in 15 children with ADHD. (11) The experimental subjects learned to inhibit the amplitude of theta waves (4–7 Hz) and increase amplitude of beta waves (15–18 Hz). Five children with ADHD were randomly assigned to a nontreatment control condition. Functional magnetic resonance imaging revealed increased activation of the right anterior cingulate cortex, an area related to selective attention that previously was shown to be altered in children with ADHD. However, it could not be determined whether the change in brain function was related to the specific neural training program (decreasing the amplitude of theta waves and increasing the amplitude of beta waves) or to the additional attentional training received by the experimental group. A 2007 report from Europe compared neurofeedback training of slow cortical potentials (SCPs) (n=17) with a control group (n=13) that participated in a group cognitive/behavior training program. (13) The report stated that randomization was incomplete because the age range in the group program was limited, parents had to be available for intense training during neurofeedback, and some parents had a preference for one type of training. Results showed that children in the neurofeedback group improved more than children who had participated in a group therapy program, particularly improved for attention and cognition. However, parental support was found to account for more of the improvement than neurofeedback training performance
To control for nonspecific effects (attention training) and confounding variables (parental engagement), Gevensleben and colleagues compared neurofeedback with a control intervention of participation in a computerized attention skills training. (12) All children were drug-naïve or drug-free without concurring psychotherapy for at least 6 weeks before starting training. The 2 training conditions were designed to be as similar as possible, using computer games, positive reinforcement by a trainer, homework, and parental encouragement in using the skills/strategies learned during training in real-life situations. Both groups participated in 2 blocks of 9 sessions (approximately 100 minutes per session plus a break), with 2–3 sessions per week, and parents were informed that both treatments were expected to be beneficial but were not informed as to which type of training their child had been assigned. A total of 102 children were randomly assigned in a 3:2 ratio; 8 children were excluded due to need for medical treatment or noncompliance with the study protocol by either the children or their parents, resulting in 59 children in neurofeedback and 35 in attention training (92% follow-up). Slow cortical potentials (SCPs) and theta/beta training were compared by starting with 1 type of training in the first block and then the other (counterbalanced order) in the second block. Investigator evaluations were performed by the teachers, and thus, the teachers were not blinded to the treatment. At the end of training/testing, there were no significant differences in parents’ attitude toward the 2 training conditions or in the perceived motivation of their children. Approximately 40% of the parents either did not know which training their child had participated in or guessed the wrong group. Both parents and teachers rated the neurofeedback group as more improved on the hyperactivity subcomponent of a Strength and Disabilities Questionnaire (e.g., SDQ, 19% vs. 3%, respectively, improved) and on a German ADHD scale (e.g., 26% vs. 9%, respectively, improved). Thirty children in the neurofeedback group (52%) and 10 children in the attention training group (29%) improved more than 25% in the German ADHD scale (odds ratio: 2.68), which was the primary outcome measure. Other components of the SDQ, including emotional symptoms; conduct problems; peer problems; and prosocial behavior, were not different between the 2 training conditions. No significant differences were noted between the 2 neurofeedback training protocols. Results of this randomized controlled study suggested that neurofeedback may have specific effects on attention and hyperactivity beyond those achieved by attention training and parental involvement. The authors concluded that future studies should further address the specificity of effects and how to optimize the benefit of neurofeedback as a treatment module for ADHD.
Six-month follow-up from the RCT described above was reported in 2010. (12,14) Of the 94 children who completed treatment, 17 started medication during the follow-up interval, and parents of 16 children did not return the questionnaires. Follow-up was obtained in 61 children (65%) of the original per-protocol 102 children. Although the percentage of dropouts did not differ between the 2 groups, dropouts tended to have higher scores on the German ADHD rating scale (FBB-HKS), particularly in the control group. The difference in dropouts between the groups limits the interpretation of the comparative data, as the scores in the 2 groups included in follow-up were not similar at baseline (e.g., baseline FBB-HKS of 1.50 for the neurofeedback group and 1.37 for the control group). The improvement observed in the neurofeedback group after treatment appeared to be preserved at 6-month follow-up. For example, the inattention subscore of the FBB-HKS improved from 2.02 to 1.51 after treatment and remained at 1.49 at 6-month follow-up (moderate effect size of 0.73). The hyperactivity/impulsivity subscore improved from 1.10 to 0.79 after treatment and remained at 0.76 at 6-month follow-up (small effect size of 0.35). The authors of this European study noted that the treatment effects appear to be limited but considered neurofeedback to be potentially effective as one component of a multimodal treatment approach.
A 2011 review of complementary medicine for ADHD indicates that there is only one large randomized trial (Gevensleben et al., reviewed above) that found a significant benefit (i.e., with a moderate effect size of 0.6) of neurofeedback for children with ADHD. (15) In comparison, effect sizes in studies that used medication were around 0.8 for methylphenidate and around 1.2 for amphetamine. Other recent studies are small and inconclusive, and larger sham-controlled studies are needed to evaluate whether neurofeedback (alone or in combination with other treatments) has beneficial effects for children with ADHD.
A 2008 systematic review of neurofeedback as a treatment for substance abuse disorders described difficulties in assessing the efficacy of this and other substance abuse treatments, including the lack of clearly established outcome measures, differing effects of the various drugs, presence of comorbid conditions, absence of a gold standard treatment, and use as an add-on to other behavioral treatment regimens. (16) The authors concluded that alpha-theta training, when combined with an inpatient rehabilitation program for alcohol dependency or stimulant abuse, would be classified as level 3 or “probably efficacious.” This level is based on beneficial effects shown in multiple observational studies, clinical studies, wait-list control studies, or within-subject or between-subject replication studies. The authors also noted that few large-scale studies of neurofeedback in addictive disorders have been reported, and a shortcoming of the evidence for alpha-theta training is that it has not been shown to be superior to sham treatment.
One small (n=6) quasi-randomized, double-blind pilot study was identified that examined whether increasing peak alpha frequency would improve cognitive performance in older adults (70–78 years of age). (17) Control subjects were trained to increase alpha amplitude or shown playback of one of the experimental subject’s sessions. Compared to controls, the experimental group showed improvements in speed of processing for 2 of 3 cognitive tasks (Stroop, Go/No-Go) and executive function in 2 tasks (Go/No-Go, n-back); other functional measures, such as memory, were decreased relative to controls.
In 2010, Cortoos et al. published a small (n=17) RCT on the effect of neurofeedback training or biofeedback training (placebo control) on objective and subjective sleep in patients with primary insomnia. (18) Of 158 subjects with sleep complaints who were interested in participating, 131 (89%) were excluded due to study criteria or unwillingness to remain medication-free during the study period. Following polysomnograph (PSG) recorded sleep in the laboratory, all subjects received 20 sessions of therapist-controlled telefeedback training at home over a period of 8 weeks. The neurofeedback group was trained to increase the sensory-motor rhythm (12-15 Hz) and inhibit theta power (4-8 Hz) and high beta power (20-30 Hz). The biofeedback group was trained to decrease electromyographic (EMG) activity, which was equated with the reinforcement of relaxation (placebo control). Both treatments reduced sleep latency by 40% to 45% (22 minutes at baseline) on post-treatment PSG, measured 2 weeks after the end of training. Neurofeedback training reduced wake after sleep onset (54% vs. 13% decrease, respectively; however, no interaction was found on the 2-way analysis of variance [ANOVA]) and increased total sleep time (40 minutes vs. less than 5 minutes, respectively, p<0.05). This study is limited by the small number of subjects, differences in sleep parameters at baseline, and short follow-up. Additional studies are needed to evaluate this novel treatment approach.
A 2009 meta-analysis by Tan and colleagues identified 63 studies on neurofeedback for treatment of epilepsy. (19) Ten of the 63 studies met inclusion criteria; 9 of these studies included fewer than 10 subjects. The studies were published between 1974 and 2001 and utilized a pre-post design in patients with epilepsy refractory to medical treatment; only one controlled study was included. The meta-analysis showed a small effect size for treatment (-0.233), with a likelihood of publication bias based on funnel plot. Randomized placebo-controlled trials are needed to evaluate the effect of neurofeedback on seizure frequency in patients with epilepsy.
A 2011 evidence review with clinical guidelines by the European Society for the Study of Tourette Syndrome identified a total of 2 case studies on neurofeedback for Tourette syndrome; this is considered experimental, investigational and unproven. (20)
Autism Spectrum Disorder
In a 2009 systematic review of novel and emerging treatments for autism spectrum disorders, neurofeedback received a grade C recommendation, supported by 1 nonrandomized controlled trial. (21) The only controlled trial identified was a pilot study from 2002 that included 12 children with autism who received neurofeedback and an untreated control group of 12 children who were matched by sex, age, and disorder severity. (22) The study found a 26% reduction in autism symptoms based on the Autism Treatment Evaluation Checklists (ATEC), compared to 3% for the untreated controls. Parental assessments found improvements in all behavioral categories (socialization, vocalization, anxiety, schoolwork, tantrums, and sleep) in the group treated with neurofeedback, while minimal changes were reported in the control group. As discussed above, there is a need for sham controlled trials with neurofeedback training due to the possibility of nonspecific effects (e.g., attention training) and confounding variables (e.g., parental engagement and expectation). No randomized sham controlled trials on neurofeedback for autism spectrum disorders have been identified.
In 2010, Kayiran and colleagues reported a randomized single blind study of neurofeedback versus escitalopram in 40 patients with fibromyalgia. (23) Patients in the neurofeedback group were instructed to widen a river on a computer monitor which corresponded to increasing sensory motor activity and decreasing theta activity. Patients received 5 sessions per week for 4 weeks. The control group received escitalopram for 8 weeks. Outcome measures at baseline and at weeks 2, 4, 8, 16, and 24 included visual analog scale (VAS) for pain, Hamilton and Beck Depression and Anxiety Inventory Scales, Fibromyalgia Impact Questionnaire and Short Form-36. Mean amplitudes of electroencephalogram (EEG) rhythms and the theta/sensory motor rhythm were also measured in the neurofeedback group. At baseline, the control group scored higher on the Hamilton and Beck Anxiety Scales and the Hamilton Depression Scale; all other baseline measures were similar between groups. Both groups showed improvements over time, with significantly better results in the neurofeedback group. There were no changes over time in mean amplitudes of EEG rhythms and essentially no change in the theta/sensory motor rhythm ratio (reduced only at week 4). This study is limited by the difference in intensity of treatment and contact with investigators between the neurofeedback and escitalopram groups. Sham controlled trials are needed when assessing the effect of neurofeedback on subjective outcome measures.
Walker reported quantitative EEG (QEEG) for the treatment of migraine headaches in 46 patients. (24) Results were compared with 25 patients who chose not to do neurofeedback and continued anti-migraine drug therapy. Since baseline QEEG assessment (all 71 patients) showed a greater amount of the high frequency beta band (21-30 Hz), the 5 neurofeedback sessions focused on increasing 10 Hz activity and decreasing 21-30 Hz targeted individually to brain areas where high frequency beta was abnormally increased. Patient diaries of headache frequency showed a reduction in migraines in a majority of patients in the QEEG group but not the drug therapy group. Fifty-four percent reported complete cessation of migraines over 1 year, with an additional 39% reporting a greater than 50% reduction. In comparison, no patients in the drug therapy group reported a cessation of headaches, and 8% had a reduction in headache frequency of greater than 50%. Randomized sham-controlled trials are needed to adequately evaluate this treatment approach.
The scientific evidence does not permit conclusions concerning the effect of neurofeedback on health outcomes; a number of questions regarding clinical efficacy remain to be answered before applying neurofeedback techniques to patients with ADHD, insomnia, epilepsy, Tourette syndrome, autism spectrum disorder, fibromyalgia, migraine headache, substance abuse disorder, or other neurologic disorders. Neurofeedback is considered experimental, investigational and unproven.
Practice Guidelines and Position Statements
The International Society for Neurofeedback & Research published a 2011 position paper on standards of practice for neurofeedback and neurotherapy. (25) Issues discussed include competency, qualifications of practitioners, scope of practice, informed consent, pretreatment assessment, standards for remote training, record keeping and billing, accountability, standards for practitioner training and qualifications to be trained, adequate supervision and coaching of training sessions, ethical advertising, standards for professional societies, and standards for those who sell and manufacture neurofeedback equipment.
Clinical guidelines on behavioral and psychosocial interventions for Tourette syndrome and other tic disorders were published in 2011 by the European Society for the Study of Tourette Syndrome. The guidelines state that neurofeedback is still experimental. (20)
The American Psychological Association (APA) provides general information on biofeedback (including neurofeedback) on their website www/apaonline.org (APA Online), stating that “Biofeedback helps treat some illness, may boost performance, helps people relax, and is even used to help children with Attention Deficit-Hyperactivity Disorder.” (26)
The Association for Applied Psychophysiology & Biofeedback (AAPB) rates neurofeedback as efficacious (level 4 on a scale of 1–5 with 5 being the best) for ADHD, based on several small controlled and moderately large clinical studies showing that neurofeedback significantly helps children with ADHD who have problems with mathematics. (27)
No information on neurofeedback was identified from the American Academy of Child and Adolescent Psychiatry, the American Psychiatric Association, or the American Academy of Pediatrics.
Disclaimer for coding information on Medical Policies
Procedure and diagnosis codes on Medical Policy documents are included only as a general reference tool for each policy. They may not be all-inclusive.
The presence or absence of procedure, service, supply, device or diagnosis codes in a Medical Policy document has no relevance for determination of benefit coverage for members or reimbursement for providers. Only the written coverage position in a medical policy should be used for such determinations.
Benefit coverage determinations based on written Medical Policy coverage positions must include review of the member’s benefit contract or Summary Plan Description (SPD) for defined coverage vs. non-coverage, benefit exclusions, and benefit limitations such as dollar or duration caps.