A Systematic Review of the Safety and Effect of Neurofeedback on Fatigue and Cognition

Abstract

Background. Many cancer survivors continue to experience ongoing symptoms, such as fatigue and cognitive impairment, which are poorly managed and have few effective, evidence-based treatment options. Neurofeedback is a noninvasive, drug-free form of brain training that may alleviate long-term symptoms reported by cancer patients. Objective. The purpose of this systematic review of the literature was to describe the effectiveness and safety of neurofeedback for managing fatigue and cognitive impairment. Methods. A systematic review of the literature was conducted using Joanna Briggs Institute (JBI) methodology. A comprehensive search of 5 databases was conducted: Medline, CINAHL, AMED, PsycInfo, and Embase. Randomized and nonrandomized controlled trials, controlled before and after studies, cohort, case control studies, and descriptive studies were included in this review. Results. Twenty-seven relevant studies were included in the critical appraisals. The quality of most studies was poor to moderate based on the JBI critical appraisal checklists. Seventeen studies were deemed of sufficient quality to be included in the review: 10 experimental studies and 7 descriptive studies. Of these, only 2 were rated as high-quality studies and the remaining were rated as moderate quality. All 17 included studies reported positive results for at least one fatigue or cognitive outcome in a variety of populations, including 1 study with breast cancer survivors. Neurofeedback interventions were well tolerated with only 3 studies reporting any side effects. Conclusions. Despite issues with methodological quality, the overall positive findings and few reported side effects suggest neurofeedback could be helpful in alleviating fatigue and cognitive impairment. Currently, there is insufficient evidence that neurofeedback is an effective therapy for management of these symptoms in cancer survivors, however, these promising results support the need for further research with this patient population. More information about which neurofeedback technologies, approaches, and protocols could be successfully used with cancer survivors and with minimal side effects is needed. This research will have significance to nurses and physicians in oncology and primary care settings who provide follow-up care and counseling to cancer survivors experiencing debilitating symptoms in order to provide information and education related to evidence-based therapy options.

Background

With aging of the population and improved cancer treatment, the number of cancer survivors is increasing. It is estimated that there are currently more than 14 million cancer survivors living in the United States and Canada.^1,2 Cancer survivors commonly experience numerous physical and psychological symptoms such as fatigue, cognitive impairment, pain, anxiety, and depression as long-term and late effects of the cancer and its treatment.^3-7 Many of these symptoms, particularly cancer fatigue and cognitive impairment, or chemobrain, are poorly managed and have few effective, evidence-based treatment options.^8-12 These issues affect the daily lives of cancer survivors, their family relationships, and return to work, often resulting in lowered quality of life.^13-15 Complementary and alternative medicine (CAM) therapy use is common among cancer patients and increasing, with cancer survivors with unmet needs more likely to use CAM to improve quality of life.^16-20

Neurofeedback is a noninvasive, drug-free form of brain training reported to help with a variety of conditions, including pain, fatigue, depression, anxiety, sleep disorders, and cognitive decline. ²¹ Thus, neurofeedback is an innovative CAM therapy with the potential to alleviate multiple long-term symptoms reported by cancer survivors and improve their quality of life.

Electroencephalography (EEG) biofeedback or neurofeedback is a scientifically based technique that can allow the brain to learn self-regulation skills. ²² These skills have clinical relevance and changes in brain regulation will be manifested in symptomatic and subjective changes. During neurofeedback, brain activity is monitored through electrodes placed on the scalp and fed back to the participant through auditory and/or visual stimuli generated through computer software without introducing anything intrusive into the brain. ²³ Rather, neurofeedback provides new information, which can open up the brain to options within its own control; if the changes result in a new stable state, or one that provides continued benefits to the participant, then it will continue to be reinforced after the session is over. ²² The mechanism of action is generally considered to be operant conditioning in which the brain is retrained to function in healthier brainwave patterns. ²⁴ The process is often described as exercise for the brain, which increases the efficiencies of specific brain function and enhances cognitive flexibility and control. ²⁴ Meditation, relaxation, martial arts, yoga, and cognitive training are other ways of gaining control over one’s mental state; however, neurofeedback is usually more rapid and efficient. ²⁵

Brain Waves and Neurofeedback Approaches

The cyclical electrical activation and deactivation of neurons in the brain can be recorded as raw unfiltered brain wave signals by EEG. ²⁶ These EEG signals can be amplified and filtered into smaller parts, called frequencies that are used during neurofeedback. Neurofeedback delivers feedback at the time of the cycle when the brain wave moves into a desirable pattern, providing simple reinforcement that teaches the brain to prolong healthy brain wave patterns. Each frequency can be measured in terms of its speed in hertz (Hz) or cycles per second, and amplitude (height of the wave) measured in microvolts (µV). Slow frequencies (<10 Hz) tend to have higher amplitudes than faster frequencies (>13 Hz) but not always. ²⁶

Single frequencies have been organized into discreet groups called bandwidths that are associated with specific characteristics: delta (1-4 Hz) associated with sleep, repair, complex problem solving; theta (4-8 Hz) associated with creativity, insight, deep states; alpha (8-12 Hz) associated with alertness, peacefulness; beta (13-21 Hz) associated with thinking, focusing, sustained attention; sensorimotor rhythm (SMR; 12-15 Hz) associated with mental alertness, physical relaxation; high beta (20-32 Hz) associated with intensity, hyperalertness, anxiety; and gamma (38-42 Hz) associated with cognitive processing and learning. ²⁶

There appears to be no single conceptual model to describe the processes and outcomes of neurofeedback among practitioners or manufacturers.^27,28 As a result, numerous nonspecific (placebo/expectancy, client/therapist/family variables/interactions), treatment-specific (treatment protocols), neurophysiological and psychological factors have been used to explain how neurofeedback is effective in the treatment of a wide variety of symptoms and disorders. ²⁹ Neurofeedback training is generally aimed at changing the amplitude (inhibition or augmentation) of selected frequency bandwidths through the principles of operant and classical conditioning; however, there are numerous approaches and protocols that can be used.

Alpha/theta training involves promotion of lower EEG frequencies for experiential and integrative purposes. ³⁰ This deep-states training is commonly used for alcoholism and posttraumatic stress disorder (PTSD). ³¹ SMR/beta training involves training to improve physiological self-regulation, and functional normalization and optimization. ³⁰ The most common neurofeedback training approaches can generally be grouped under 3 guiding philosophies. (1) Targeted mechanisms-based, symptom-responsive approaches in which the neurofeedback provider selects a reinforcement brainwave target for the specific symptom are typically applied to traditional psychopathologies such as anxiety, depression and bipolar disorders, attention disorders, personality disorders, trauma recovery, addictions, pain syndromes, sleep disorders, and specific learning disabilities ³² ; (2) QEEG-driven approaches that seek EEG normalization based on stationary QEEG measures are typically applied to more severely affected clinical populations such as neurological problems, traumatic brain injuries, strokes, seizures, dementias, developmental delay, cerebral palsy, and autism; ³² and (3) nonlinear dynamical (NLD) approaches that are generalized approaches to neuroregulation based on the theory of nonlinear dynamical systems ²⁷ are generally applied to nonclinical populations for optimum mental fitness, brain exercise and meditation practice. ³⁰ However, all 3 approaches can be effective in all these domains. ³²

Other nontraditional neurofeedback approaches can also be used. A low-energy neurofeedback system (LENS) approach uses weak electromagnetic signals to provide feedback to the brain. ³³ Like NLD neurofeedback, LENS use a nonspecific treatment approach that is not guided by diagnosis.^33,34 Hemoencephalography (HEG) neurofeedback or cerebral flow neurofeedback, is used to train individuals to improve cerebral blood flow to key areas of the brain which improves cerebral oxygenation and translates into enhanced mental capacity. ³⁵ HEG uses physiological signals based on blood flow dynamics rather than electrical activity and is reported to be particularly helpful with head injuries, attention deficit disorder and migraines.^35,36

Clinical Applications of Neurofeedback

Biofeedback has recently been recognized by the American Academy of Pediatrics as a “Level 1—Best Support” intervention for attention and hyperactivity behaviors. ³⁷ Neurofeedback is also reported to help with a variety of conditions, including chronic pain, chronic fatigue syndrome, depression, anxiety disorders, sleep disorders, PTSD, migraines, and cognitive decline. ²¹ It is unclear whether cancer survivors experiencing similar symptoms would benefit from neurofeedback and which approaches or protocols would be most effective for managing cancer fatigue and cognitive impairment.

In the Canadian Cancer Society pamphlet Complementary Therapies: A Guide for People With Cancer, biofeedback is described as a type of mind-body therapy used to improve quality of life. ³⁸ In a clinical practice guideline for integrative oncology, mind-body modalities of CAM therapies are recommended for reducing anxiety, mood disturbance, and chronic pain and for improving quality of life in cancer patients; these techniques include meditation, hypnosis, relaxation, cognitive-behavioral therapy, biofeedback, and guided imagery. ³⁹ A published study protocol describes a clinical trial that will investigate the effectiveness of neurofeedback using SMR/beta training to improve neurocognitive functioning of children treated for a brain tumor. ⁴⁰

Results from a recent prospective study by Alvarez et al ⁴¹ demonstrated the feasibility of neurofeedback for breast cancer survivors. The sample initially demonstrated significant dysfunction compared with general population norms prior to neurofeedback, and strongly significant improvements (P < .001) postneurofeedback on all 4 cognitive measures, a fatigue scale, 4 psychological scales, and 3 of 8 sleep scales; improvements were generally linear across the course of training, and were maintained at follow-up testing. ⁴¹ These results suggested neurofeedback deserved further study as a novel, method of addressing long-term symptoms in cancer survivors that may be safe, effective, and acceptable to patients.

Unfortunately, neurofeedback is not widely accepted by the general public or medical community because of lack of information and insufficient research to demonstrate its benefits, and lack of coverage by public or private health insurance plans. ⁴² The Canadian Integrative Oncology Research Initiative identified 5 interrelated priority research areas, including clinical effectiveness and safety of complementary therapies. ⁴³ Information about which neurofeedback technologies and protocols are being used successfully and with minimal side effects is needed.

Purpose

The purpose of this review was to explore the effect and safety of EEG biofeedback or neurofeedback on fatigue and cognitive impairment which are long-term symptoms commonly experienced by cancer survivors. The review attempted to answer the following research questions:

1. What is the effect of neurofeedback on fatigue and cognition?

2. What side effects and/or adverse events are experienced during or following neurofeedback therapy administered for management of fatigue or enhancing cognition?

3. Which neurofeedback systems, approaches and/or protocols have demonstrated effectiveness for management of fatigue and/or enhancing cognition?

Methods

Data Sources

This systematic review was conducted using Joanna Briggs Institute (JBI) methodology, including JBI critical appraisal and data extraction tools. ⁴⁴ The research team had experience in review methodologies and was supported by a library scientist. Inclusion criteria were defined a priori. A comprehensive systematic search of 5 databases was conducted: OVID MEDLINE (1946 to November week 3, 2013); EMBASE (1947 to 2013 week 44); PsycINFO (1967 to December week 1, 2013); AMED (1985 to December 2013); and CIHAHL (1978 to December week 1, 2013). Hand searches were also conducted of the major journals reporting on neurofeedback and reference lists of included studies. The search strategy in the MEDLINE database was as follows (numbers in parentheses indicate number of articles at each stage of the search):

1. Neurofeedback/(232)

2. neurofeedback.mp. (453)

3. brain biofeedback.mp (1)

4. eeg biofeedback.mp (105)

5. 1 or 2 or 3 or 4 (523)

6. exp Fatigue/(20 824)

7. exp Cognition disorders/(64 144)

8. exp Cognition/(111 419)

9. 6 or 7 or 8 (189 112)

10. 5 and 9 (56)

A study was eligible for inclusion in this review if it (1) reported on adult cancer survivors, individuals with other chronic health conditions or nonclinical populations seeking to decrease fatigue and/or enhance cognitive abilities; (2) reported results of randomized controlled trials, nonrandomized controlled trials, controlled before and after studies, cohort, case control, or descriptive studies that assessed effectiveness and/or safety of EEG biofeedback or neurofeedback therapy; and (3) was written in English. We excluded studies on children, editorials, reviews, expert opinion papers, and studies published as abstracts only.

Study Selection and Critical Appraisal

Titles and abstracts were screened to determine broad eligibility using the inclusion and exclusion criteria. All citations identified as “include for critical appraisal” or “unclear” were rescreened using full text reports. Papers selected for critical appraisal were assessed by the principal investigator (MLF) and another reviewer (DG or LP) for methodological quality prior to final inclusion in the review. One of 3 JBI standardized critical appraisal instruments were used depending on the study design: (1) JBI Critical Appraisal Checklist for Experimental Studies ⁴⁵ ; (2) JBI Critical Appraisal Checklist for Comparable Cohort/Case Control Studies ⁴⁶ ; or (3) JBI Critical Appraisal Checklist for Descriptive/Case Series. ⁴⁷ Given the paucity of research in this area, a cutoff score of 4 was established for each JBI checklist to ensure a sufficient number of studies to review while maintaining the strength of the methodological quality. A score of 4 to 6 indicated moderate quality, whereas as score of 7 or more indicated high quality. See the diagram in Figure 1 for a summary of the search decision flow.

Figure 1.

Search decision flow diagram.

Data Extraction, Synthesis, and Analysis

The primary reviewer (MLF) extracted data from the 17 included studies using a modified JBI data extraction form to summarize the pertinent details from each study. Extracted data were transferred to synopsis tables to synthesize relevant data. Because this systematic review was conducted using JBI methodology, the results were presented in separate tables by study design (ie, experimental vs descriptive studies). The data extracted included the following specific details of significance to the review questions and objectives: reference, study design, patient population/sample, neurofeedback intervention (system, protocols, number of sessions), outcome measures, efficacy results, safety results, limitations, conclusions, study quality. The primary outcomes of interest were levels of chronic fatigue and cognitive impairment. Secondary outcomes of interest were side effects and adverse events. Given the heterogeneity of the intervention systems and protocols, no meta-analysis was conducted.

Results

Quality Assessment

Twenty-seven relevant studies were included in the critical appraisals. The quality of most studies was poor to moderate based on the JBI critical appraisal checklists. Seventeen studies were deemed of sufficient quality to be included in the review. Of these, only 2 were rated as high-quality studies,^45,46 and the remaining were rated as moderate quality.^47-60

Findings

Ten experimental studies (see Table 1) and 7 descriptive studies (see Table 2) were synthesized in this review.

Table 1.

Data Extraction Experimental Studies.

Author/Date/Country	Study Design	Patient Population/Sample	Neurofeedback Intervention	Outcome Measures	Effectiveness Results	Safety Results	Limitations	Conclusions	Study Quality
1. Angelakis et al, 2007 USA	Quasi-experimental	Healthy elderly n = 4 women n = 2 men 70-78 y n = 3 Exp (Experimental) n = 2 Cont (control) n = 1 Pseudo	Parietal-occipital midline electrode POz BioGraph/ProComp Peak alpha frequency (PAF) 1-2/wk 31-36 sessions versus alpha amplitude neurofeedback (NF) and pseudo NF	Cognition: Self-rated standard psychometric tests; “n-back” and “Go/No-Go” oddball cognitive tasks	• PAF group self-rated as thinking faster but no change in concentration • PAF group no change for verbal memory, worse for visual memory, whereas active controls improved visual memory • PAF group increased response speed and accuracy on Stroop test and improved reaction time on “n-back” test	• Some protocols may worsen some performance items	• Small sample size, serve as pilot data only • Participants were high-motivated and high-functioning seniors	• PAF NF improved cognitive processing speed and executive function No clear effect on memory PAF NF showed cognitive improvements in areas that alpha amplitude NF (control) did not and vice versa • Supports idea that EEG biofeedback is protocol-specific resulting in different effects depending on the frequency and modality of training	Moderate
2. Becerra et al, 2012 Mexico	Randomized controlled trial (RCT)	Normal elderly 60-84 y n = 9 female n = 5 male n = 7 Exp 65.8 ± 2.4 y n = 7 Cont 67 ± 4.9 y	MEDICID IV Reduce theta absolute power (AP) 30 sessions Over 10-12 wk Reinforcement administered intermittently (60% to 80% of the time) versus sham NF	Cognition: Wechsler Adult Intelligence Scale (WAIS-III) and NEUROPSI attention, executive function and memory	• Verbal Comprehension Index (P = .02) and Verbal IQ (P = .05) from WAIS-III and Total Score on the NEUROPSI test (P = .01) increased significantly for NF group • In both groups, Memory Score of the NEUROPSI increased significantly (P = .01) • In control group, there was a significant decrease in the WAIS Performance IQ (P = .05) • Reduced theta AP and increased alpha AP were only observed in the NF group		• Small sample size, used nonparametric permutational analysis of variance (ANOVA) model with 2 factors (group and time)	• Experimental NF group exhibited greater improvement in EEG and behavioral measures; however, subjects of the control group also showed improved EEG values and in memory, which may be attributed to a placebo effect• Effect of the NFB treatment was clear in the experimental group, although a placebo effect may also have been present	Moderate
3. Fritson et al, 2007USA	RCT	Nonclinical psychology college studentsn = 3219-38 yn = 28 femalen = 4 malen = 16 Expn = 16 Cont	Beta/sensorimotor rhythm (SMR)C3 (active)-C4 (referential) bipolar sensor placement with A1 ground sensor20 sessions2/wkversus sham NF	Attention: Integrated Variables of Attention + Plus (IVA + Plus); Kaufman Brief Intelligence Test 2 (K-BIT 2); Wide Range Intelligence Test (WRIT)Fatigue	• Multivariate analysis of variance (MANOVA) found significant main effect of training (P < .15) for the IVA Full-Scale Response Quotient (FSRQ); main effect of group and interaction were not significantAttention and intelligence scores did not show significant change		• •Participants only completed 20 sessions, would have preferred 30-40 sessions• Originally enrolled 39, but 7 dropped out because of scheduling conflicts• Floor effects may have inhibited chance of finding significant changes for intelligence	• SMR EEG biofeedback may affect some characteristics of nonclinical individuals in as few as 20 sessions• Specifically EEG biofeedback improved response control, including responses to auditory and visual stimuliProvides empirical support for use of SMR EEG biofeedback with healthy adults to improve response control (improve response accuracy, decrease impulsivity and decrease fatigue) even if there is no obvious impairment, which is consistent with the idea of peak performance• A greater number of sessions may be necessary for attention processes of healthy adults to be impacted by EEG biofeedback	Moderate
4. Hoedlmoser et al, 2008Austria	RCT	Healthy individualsn = 27n = 13 malen = 14 femaleMean age 23.63 yn = 16 Expn = 11 Cont	THERA PRAXSMR conditioning: instrumental conditioning of sensorimotor rhythm (ISC) protocol10 sessionsversus randomized-frequency conditioning	Declarative memory: Declarative word-pair association taskMemory: Wechsler Memory Scale-RevisedIntelligence: Advanced Progressive Matrices	• Calculated a 2-way ANOVA for retrieval scores (declarative memory), which increased after conditioning (P = .004) but no overall group difference• Only SMR group had higher retrieval scores after 10 sessions (P = .018)			• Significant SMR amplitude changes from early to late conditioning sessions confirmed success of the ISC• These EEG changes transferred into sleep and even improved immediate memory retrieval after learning• Memory consolidation was unchanged• Possibly heightened attention or relaxation levels (during learning and/or retrieval) after 10 ISC sessions, thus account for the observed improvement in word-pair recall	Moderate
5. Kayiran et al, 2010Turkey	RCT	Fibromyalgian = 36n = 18 ExpMean age 31.78 yn = 18 ContMean age 32.39 yversus escitalopram	Alien Technik3/32 and Brainfeedback-3SMR20 sessionsOver 4 wk	Fatigue: Visual analogue scale (VAS) for fatigueFibromyalgia Impact Questionnaire (FIQ)	• VAS-fatigue levels scores decreased significantly in both groups (P < .001)• Mean VAS-fatigue scores were lower than those of the control group in every visit during follow-up (P < .05)• A significant decrease in theta/SMR ratio was observed at the end of treatment• The therapeutic efficacy of NF treatment began at second week and reached maximum effect at fourth week• Effect of selective serotonin reuptake inhibitor (SSRI) treatment began at second week but reached maximum effect at eighth week		• 2 patients in the NF group and 2 in the control group were lost to follow-up• Small sample size• Lack of any process which can lead to explain the mechanism of NF for FMS	• All posttreatment measures showed significant improvements in both groups• NF group displayed greater benefits than controls• Data support the efficacy of NF as a treatment for pain, psychological symptoms, and impaired quality of life (QOL) associated with fibromyalgia	Moderate
6. Keller, 2001Germany	Quasi-experimental	Traumatic brain injuryn = 2121-42 yn = 12 Expn = 9 Cont	Beta activity training10 sessionsOver 2 wkversus computer-based attention training	Attention	• NFT group increased the duration they were able to sustain beta activity above the threshold (P = .012)• NFT group improved more than controls in the sustained attention task• Control group improved mainly in computerized attention tests requiring focused attention for short periods of time• NFT group improved in computer-based and paper-pencil tasks			• NFT had a more extensive effect on attention deficits than the computer-based therapy• suggests that NFT should focus not only on enhancement of beta activity but also on duration patients are able to hold beta activity	Moderate
7. Kravitz et al, 2006USA	RCT	Fibromyalgian = 64n = 33 Expn = 31 ContMean age 46.9 yn = 59 femalen = 5 male	Flexyx neurotherapy system (FNS)Low-energy neurofeedback system (LENS)22 sessionsOver 11 or more weeksversus sham	“Fibro-fog”CognitiveFatigueFibromyalgia Symptom Scales;Clinical Global Impressions Scales: Clinician (CGI-I) and participant (PGI-I); Fibromyalgia Impact Questionnaire (FIQ); CNS Dysfunction Questionnaire	• More participants treated with FNS than with sham improved partially or fully on the CGI-I at session 22 (P = .01) and follow-up (P = .04)• There was no significant treatment effect on any secondary outcome measure and no specific symptom improved preferentially with active versus sham FNS	• Side effects monitored at each session graded 0-3• 52.5% reported at least one side effect at any time during treatment• Side effects differed significantly (P < .007) between active (74.2%) and sham (35.7%)• Symptom reported most commonly fatigue was reported by 13 (10 in Exp); pain, including headache was reported by 10 (6 in Exp); 4 Exp reported sleep problems; 3 in Exp reported stiffness or muscle spasms• None dropped out due to side effects, but in a few cases, treatment was suspended temporarily	• Those with debilitating chronic fatigue were not included; thus sample may have been atypical of FMS patients seeking treatment	• Only the clinician-rated global impressions scores detected a treatment-related response, which persisted through 1 week posttreatment; significantly more participants treated with FMS were rated as partially or fully remitted• A pretreatment delta/alpha EEG amplitude ratio >1 was associated with PGI-I but not CGI-I response independent of treatment group assignment• Participants in the study FMS group were more likely to rate study treatment as more effective than previous treatment, and sham group were more likely to rate study treatment as no different or worse than previous treatment• Most important finding may be that for FMS patients EEG treatment alone is not sufficient for recovery	High
8. Lecomte and Juhel, 2011France	Quasi-experimental	Elderlyn = 3065-85 yMean age 75.25 yn = 23 womenn = 7 men	Pendant EEG system/BioExplorerTransverse captors:C3-Cz/Cz-C4Alpha/theta training4 sessions1/wkversus Relaxation versus Test/retest	Memory: 2 of 3 memory tests from the Signoret Memory Battery	• Overall, an improvement of memory performance was observed in members of all 3 groups (P = .001)• Memory performance of several members of the NFB group increased significantly more than the average increase of memory performance observed in members of the TRT group• Increased performance in both memory tests was only significant in one member of the NFB group• No relation was observed between changes in EEG activity and improved performance in either memory test		• Relatively small sample size• Few training sessions• Overall good level and relative homogeneity of memory performance displayed by participants• Type of task used to assess memory performance may have been insufficiently matched with the cognitive process associate with the frequency of wave ranges targeted by NFB training	• NFB training resulted on average in an increase of the spectral power of the upper alpha waves and in an increase of the alpha/theta ratio, despite the limited number of training sessions• The improved memory performance observed in several members of the NFB group cannot be attributed to changes in the spectral power of upper alpha waves or of theta waves, which NFB training might have enabled• The number of training sessions is probably insufficient	Moderate
9. Raymond et al, 2005England	RCT	Medical students with high withdrawal scoresn = 12NF 24.3 yMock 19.83 y	BrainmasterAlpha/theta training10 sessions2/wkversus mock NF	FatigueEnergy: Profile of Mood States (POMS)	• Theta/alpha ratios rose significantly with time within sessions (P < .01) but did not change significantly across sessions• Sessions of real NF produced on average greater improvements in mood than did sessions of mock feedback• Changes were most notable in the subscale “energized-tired” NF group feeling much more energized, and mock group feeling less energized			• Results suggest that alpha/theta NF is a mood-enhancing and energizing experience that is not attributable to relaxation or pleasant sounds alone as heard in the mock condition	High
10. Schoenberger et al, 2001USA	RCT (wait list control)	Traumatic brain injury (TBI)n = 1221-53 yn = 2 menn = 12 women	J & J Enterprises 1-400 EEG biofeedbackFlexyx Neurotherapy System (FNS)25 sessionsOver 5-8 wkversus 6- to 8-week wait-list control	Fatigue: Individualized Symptom Rating Scale; Multidimensional Fatigue Inventory (MFI)Cognition: Auditory Verbal Learning Test (AVLT); Paced Auditory Serial Addition Test (PASAT); Rey-Osterrieth Complex Figure; Trail Making Test; Controlled Oral Word Association; Digit Span Backwards; Digit Symbol	Between-group analyses:• Treatment group was significantly improved compared with the control group on the individualized rating scale (P < .01)• Between-group differences were not significant for the total MFI score but significantly improved on the General Fatigue and Mental Fatigue subscales (P < .02)• Treatment group was significantly improved compared with the control group on Digit Span Backwards (P < .05), the interference trial (P < .05) and the delayed recall trial (/<.03) of the AVLT and the most difficult trial of the PASAT (P < .02)• Within-group analyses:• Showed significant improvement over time for each self-report measure except the reduced activity and reduced motivation subscales of the Fatigue Inventory; among the fatigue subscales, participants reported the greatest improvement in their mental fatigue• Most neuropsychological measures showed significant improvements over time• Before treatment, all 12 participants reported difficulty with their ability to work or complete academic courses; after treatment, 7 were able to work professionally or engage in full academic work• With treatment, 2 of 8 participants who had been taking medications were able to reduce the dosage and 3 were able to eliminate them entirely	• Most common side effect experienced during treatment was a temporary intensification of symptoms that had previously been problematic, including pressure in the head or headache (3 people), dizziness (4 people), nausea (1 person), tingling sensation (1 person) or physical pain associated with injuries at the time of the accident (2 people)• Such reactions usually occurred within the first 6 or 7 sessions and typically resolved within a few days• Adverse reactions may also be caused by overtreatment; at each session participants were asked about their reactions to treatment to determine the length of the next treatment session; fatigue (3 people) and restlessness (1 person) were the most common indications of overtreatment and session length was reduced accordingly; following treatment 1 person reported unexplained hair loss	• Sample size may have been too small to detect an effect on some measures	• The overall pattern of results from this preliminary investigation strongly suggests that FNS may be an efficacious treatment for people who have experienced a TBI• Compared with a wait-list control FNS treatment produced a significant improvement in depression and a range of other symptoms reported by participants as most problematic• Treatment also produced significant improvements on some measures of cognitive function, specifically those involving working memory, immediate memory of new material and retention of information; results for other measures also indicated improvement but failed to reach statistical significance• Follow-up assessments showed clearly that improvements were maintained after the end of treatment	Moderate

Table 2.

Data Extraction Descriptive Studies.

Author/Date/Country	Study Design	Patient Population/Sample	Neurofeedback Intervention	Outcome Measures	Efficacy Results	Safety Results	Limitations	Conclusions	Study Quality
1. Alvarez et al, 2013USA	Pre-post with wait-list control period	Breast cancer survivorsn = 2343-70 y	NeurOptimalNonlinear dynamical (NLD) approachSensors at C3, C42×/wk20 sessions	Cognition: Functional Assessment of Cancer Therapy-Cognitive Function (FACT–Cog)Fatigue: Functional Assessment of Chronic Illness Therapy—Fatigue (FACIT-Fatigue)	• Initially the sample showed serious dysfunction on all measures compared with population norms• At posttest, the sample no longer differed significantly from the normative populations on 3 of 4 FACT-Cog measures and the FACIT-Fatigue• Repeated-measures ANOVAs revealed strongly significant improvements (P < .001) on all 4 cognitive measures and the fatigue scale	• NF was well tolerated, when questioned before and after each session none of the participants reported any side effects or other adverse events	• Sample was small, middle-aged and exclusively Caucasian and female• Baseline cognitive measures were unavailable• Serial neuropsychological testing and functional neuroimaging were not performed	• Breast cancer survivors showed significant baseline impairments in self-reported cognitive function and fatigue as compared with a normal population; after 20 sessions of NF, their performance in these areas had improved to levels indistinguishable from population norms• There were no dropouts from the study• Does not provide any insight into specific mechanisms by which EEG biofeedback might exert an effect on cognitive function	Moderate
2. Caro and Winter, 2011USA	Case-control	Fibromyalgian = 15 EEG-BF FMS patientsn = 14 femaleMean age 66.7 ± 12.3 yn = 63 Contn = 50 femaleMean age 50.5 ± 13.9 y	NeurocyberneticsSMR augmentTheta inhibitionHigh beta40-98 sessions	Attention: Continuous performance test (CPT) test of attention variablesFatigue: subjective 0-10 scale	• Mean symptom score for fatigue improved by 40% (P = .006) at completion of EEG BF• 4 of 6 CPT subtest scores showed significant improvement compared with baseline (P < .01)			• Attention abnormalities exist in FMS and certain physical abnormalities in FMS improve with EEG-BF• Attention measuring tests such as the computer based CPT used here may be useful modalities for following FMS patients over time and during interventional EEG-BF studies and may help avoid examiner bias in following FMS subjects’ response to therapy	Moderate
3. Larsen et al, 2006USA	Uncontrolled case series, retrospective	Central nervous system (CNS) problemsn = 1006-80 y	Low-Energy Neurofeedback System (LENS)Positive offset frequency10+ sessions	Cognition and Fatigue: Subjective Symptom Rating Scale (SSRS)	• Every symptom category decreased over the course of treatment, demonstrating that LENS treatment was clinically effective in ameliorating widely diverse CNS-related problems• The decline in the average ratings of symptoms was found to be greater in early sessions of treatment with further improvements occurring at a more gradual pace over time• the r-squared values in rank order of improvement for symptom categories were: disorganization, .985; cognitive problems, .983; attention, .956; fatigue, .955; mood disturbances .954; pain, .941; anxiety, .928; executive function, .903; miscellaneous problems, .894; sleep disturbances, .891; somatic complaints, .874; flexibility, .864; behavioral problems, .857; dissociation, .715; and addictions, .0003		• Confounding factors: about half of patients were on prescription medications	• Mean symptom ratings at post-treatment were exactly half of the pretreatment symptom levels (P < .0001)Overall EEG amplitudes were found to significantly (P < .0001) decline over the course of treatments at the highest amplitude site (HAS)EEG improvement was also highly correlated (r 2 = .869) with improvements in symptom ratings• LENS provides a very encouraging therapeutic option to traditional neurofeedback for the treatment of a wide range of clinical, brain-related conditions, particularly because LENS requires minimal cooperation and allows the patient to remain passive	Moderate
4. Thornton, 2000USA	Case series	Brain injuryn = 5n = 3 malen = 2 femalen = 2 normaln = 2 brain injuryn = 1 surgery for seizure control11-50 y	Lexicor Medical Technology QEEG17-36 sessions	Memory: Story recall	• Improvements ranged from 68% to 181% in the patients with brain injury				Moderate
5. Thornton, 2002USA	Case series	Brain injuryn = 5n = 1 malen = 4 femalen = 1 normaln = 3 brain injuredn = 1 frontal hematoma29-69 y	Lexicor Medical TechnologyQEEG13-55 sessions	Memory: Paragraph recall	• Improvements ranged from 39% to 134% and either maintained or improved in all of the subjects who had a follow-up assessment 1 month to 1 year following end of treatment			• Results suggest that electrophysiological variables can be changed by EEG biofeedback (BF)EEG BF can be an effective tool in the rehabilitation of the auditory memory in persons with brain damage and normal persons• There is a meaningful relationship between the electrophysiological variables and memory functioning, which can be addressed effectively with EEG BF	Moderate
6. Thornton and Carmody, 2013USA	Case series	Traumatic brain injury; learning disability; normaln = 597-65 yn = 21 femalen = 38 malen = 15 TBIn = 12 normaln = 17 adult SLD (specific learning disabilities)n = 15 child SLD	Lexicor Medical TechnologyQEEGM =45 sessions	Memory	• Analysis revealed significant improvements in auditory and reading memory across all groups as well as changes on the QEEG variables• All the groups were performing above the normative reference group on measures of auditory and reading memory (24% to 97%)Average auditory memory improvement was 82%Average reading task memory improvement was 154%			• The findings of significant changes in the Spectral correlation coefficient values must be presumed to underlie much of the improved cognitive functions	Moderate
7. Walker et al, 2002USA	Pre-post	Mild closed head injury (MHI)n = 26n = 12 malen = 14 female25-65 y	EEG Spectrum Neurocybernetics 3.1QEEGCoherence training5-41 sessions	Memory: Global Improvement Score (GIS)	• Significant improvement (>50%) was noted in 88% of the patients (mean 72.7%)all patients reported they were able to return to work following the treatment			• In this uncontrolled open trial of QEEG-guided coherence training, the majority of patients with MHI experienced substantial and rapid symptomatic improvement, including return to work	Moderate

Characteristics of the Included Studies

The 17 included studies were published in English between 2000 and 2013 in 7 countries. Ten of the studies were conducted in the United States.^{41,45,47,49,54-59} and 1 each from Mexico, ⁴⁸ Austria, ⁵⁰ Turkey, ⁵¹ Germany, ⁵² France, ⁵³ and England. ⁴⁶ Seven of the 10 experimental studies were randomized controlled trials,^{45,46,48 -51,54} and 3 had quasi-experimental study designs.^47,52,53 The descriptive studies consisted of 4 uncontrolled case series,^56-59 2 pre-post studies,^41,60 and 1 case-control study. ⁵⁵

Six studies reported on results with patients with traumatic brain injuries,^52,54,57-60 3 with fibromyalgia,^45,51,55 1 with breast cancer survivors, ⁴¹ 1 with patients with central nervous system (CNS) problems,^41,55-60 and 1 with medical students having high withdrawal scores. ⁴⁶ Additionally, 5 studies reported on results with nonclinical populations. Three studies were conducted with healthy elderly persons,^47,48,53 1 with healthy adults, ⁵⁰ and 1 with healthy college students. ⁴⁹ Sample sizes ranged from 6 to 64 for the experimental studies and from 5 to 100 for the descriptive studies. Participant ages ranged from 19 to 84 years for the experimental studies and from 6 to 80 years for the descriptive studies.

Neurofeedback Interventions Used in the Included Studies

A wide variety of neurofeedback systems and technology were used in the included studies, however they can be grouped into 5 general types of neurofeedback training. Four experimental studies used alpha/theta training^46-48,53; 4 experimental studies^49-52 and 1 observational study ⁵⁵ involved SMR/beta training; 4 descriptive studies involved QEEG training^57-60; 1 descriptive study employed an NLD approach ⁴¹ ; and 2 experimental studies^45,54 and 1 descriptive study ⁵⁶ used a LENS approach. None of the studies reported on HEG biofeedback. Interventions ranged from 4 to 98 neurofeedback sessions 1 to 5 times per week, with most studies reporting from 10 to 40 sessions delivered once or twice per week.

Effect on Fatigue and Cognition

All the included studies reported positive results for at least 1 fatigue or cognitive outcome as measured by a variety of standard self-rating questionnaires and cognitive task tests. Fourteen of the studies reported improvements in cognition,^{41,45,47-49,52-60} and 8 of the studies reported improvements in fatigue.^{41,46,49-51,54-56}

Alpha/theta training demonstrated mixed results on cognitive performance in three experimental studies with healthy elderly subjects. In the pilot study by Angelakis et al, ⁴⁷ a peak alpha frequency (PAF) neurofeedback protocol resulted in improvement in speed of processing and executive functioning but had no clear effect on memory, whereas an alpha magnitude protocol showed improvement in memory. In the study by Becerra et al, ⁴⁸ healthy seniors improved on 2 measures from the Wechsler Adult Intelligence Scale (WAIS-III), Verbal Comprehension (P = .02) and Verbal IQ (P = .05) as well as the total score on the NEUROPSI neuropsychological test battery (P = .01), which evaluates attention, executive function, and memory. The Memory Score of the NEUROPSI increased significantly in both the experimental and control groups (P = .01). Improved memory performance (P = .01) as measured by the Signoret Memory Battery was observed in members of all 3 treatment groups in the study by Lecomte and Juhel, ⁵³ which compared neurofeedback (NFB) to relaxation and a test/retest (TRT) group. Memory performance of several NFB participants improved significantly more than the average increase of increase observed in the TRT group (P = .001 to P = .022). Raymond et al ⁴⁶ reported that medical students with high withdrawal scores given reported feeling more energetic (P < .01) given alpha/theta neurofeedback, and that participants given mock feedback felt more tired (P < .05).

Sensorimotor rhythm/beta training resulted in a significant main effect (P < .05) among a sample of nonclinical psychology students for the Integrated Variables of Attention Full-Scale Response Quotient (IVA FSRQ), which measures response control (response accuracy, impulsivity, and fatigue). ⁴⁹ Instrumental conditioning of SMR resulted in improved declarative memory performance in a study with healthy individuals as evidenced by higher retrieval scores on the Wechsler Memory Scale–Revised after only 10 sessions (P = .018). ⁵⁰ Two studies reported on the use of SMR training for individuals with fibromyalgia. Kayiran et al ⁵¹ reported significantly lower VAS-fatigue scores following SMR training versus patients treated with escitalopram. Fibromyalgia patients in the study by Caro and Winter ⁵⁵ reported 40% improvements in fatigue as measured by a subjective 0 to 10 scale (P = .006) and significant improvements in attention on 4 of 6 Continuous Performance Test (CPT) subscores (P < .01). A sample of patients with traumatic brain injury improved more in a sustained attention task following beta activity training than a control group who received computer-based attention training (P = .006).

Quantitative EEG training improved cognitive measures in 4 studies of patients with brain injuries. Two case series studies by Thornton^57,58 reported improvements in memory ranging from 39% to 181% as measured by story/paragraph recall. A third, larger study by Thornton and Carmody ⁵⁹ recorded 105% improvements in auditory memory and 143% improvements in reading memory among the traumatic brain injury subjects. Significant memory improvement (>50%) was noted in 88% of mild closed head injury patients as measured by the Global Improvement Score (GIS) in a pre-post study by Walker et al. ⁶⁰ All the patients in the study reported being able to return to work following the QEEG neurofeedback therapy.

Nonlinear dynamical training was reported in only 1 study, a pre-post study with a wait-list control in which Alvarez et al ⁴¹ demonstrated significant improvements in both cognition and fatigue in a sample of breast cancer survivors experiencing serious cognitive impairment and fatigue following cancer treatment. At posttest, the sample no longer differed significantly from the normative population on 3 of 4 Functional Assessment of Cancer Therapy–Cognitive (FACT-Cog) measures and the Functional Assessment of Chronic Illness Therapy–Fatigue (FACIT-Fatigue) scales. Subjects demonstrated significant improvements on all 4 cognitive measures and the fatigue scale after only 20 sessions.

Low-energy neurofeedback system training was effective in a randomized controlled trial of 64 persons with fibromyalgia as rated by clinicians on the Clinical Global Impressions Scale–Clinician (CGI-I) at session 22 (P = 0.01) and follow-up (P = .04), but there was no significant effect on the participant version of the scale (PGI-I) or the Cognitive Fatigue Fibromyalgia Symptom Scales or the Fibromyalgia Impact Questionnaire (FIQ) or the CNS Dysfunction Questionnaire. ⁴⁵ LENS was also used in a smaller trial with traumatic brain injury patients. ⁵⁴ The therapy resulted in significant improvements on some measures of cognitive function involving working memory versus a 6- to 8-week wait-list control group: Digit Span Backwards (P < .05), Interference Trial (P < .05), Delayed Recall Trial (P < .03) of the Auditory Verbal Learning Test (AVLT), and the most difficult trial of the Paced Auditory Serial Addition Test (PASAT; P < .02). Fatigue scores also improved significantly as measured by the Individualized Symptom Rating Scale (P < .01) and the General Fatigue and Mental Fatigue subscales (P < .02) of the Multidimensional Fatigue Inventory (MFI) in comparison with the control group. An uncontrolled case series of 100 subjects with CNS problems demonstrated that LENS therapy was clinically effective with mean symptom ratings on the Subjective Symptom Rating Scale (SSRS) at posttreatment falling to exactly half of the pretreatment levels (P < .001). ⁶¹ The top four r ² values in rank order of improvement were disorganization (.985), cognitive problems (.983), attention (.956), and fatigue (.955). Executive function ranked eighth (.903).

Safety of Neurofeedback

Only 4 of the included studies commented on the side effects or safety of the neurofeedback interventions. Angelakis et al ⁴⁷ indicated that some neurofeedback protocols may worsen some performance items. In the RCT by Kravitz et al, ⁴⁵ side effects were monitored at each session, with 52.5% of subjects reporting at least 1 side effect at any time during the treatment. Side effects differed significantly (P < .007) between the LENS treatment group (74.2%) and the control group (35.7%) with the most commonly reported symptoms being fatigue, pain, sleep problems, and stiffness and muscle spasms. No subjects dropped out of the study due to side effects, however, in a few cases the therapy was temporarily suspended. The most common side effect experienced in the LENS trial by Schoenberger et al ⁵⁴ was a temporary intensification of previously problematic symptoms such as pressure in the head or headache, dizziness, nausea, tingling sensation, or physical pain associated with injuries. These reactions typically occurred within the first six or seven sessions and resolved within a few days. Fatigue and restlessness were the most common indications of overtreatment and session length was reduced accordingly. And finally, in the pre-post trial of breast cancer survivors conducted by Alvarez et al, ⁴¹ the NLD neurofeedback approach was well tolerated and none of the participants reported any side effects or adverse effects.

Discussion

The aim of this review was 3-fold: (1) to describe the effect of neurofeedback on fatigue and cognition, (2) to describe side effects and/or adverse events experienced during or following neurofeedback therapy administered for management of fatigue or enhancing cognition, and (3) to describe neurofeedback systems, approaches and/or protocols that have demonstrated effectiveness for management of fatigue and/or enhancing cognition.

Research dating back to the 1940s has demonstrated that brainwave or EEG activity could be trained through classical and operant conditioning using biofeedback techniques. ⁶² Clinical applications have ensued in the fields of neurology and psychology, protocols and equipment have been refined, and evidence has been building for its use in treatment of epilepsy, attention deficit disorder, closed head injury, learning disabilities, anxiety, depression, alcoholism and other addictions and autism spectrum disorders. ⁶³ Research studies have also reported that neurofeedback helps with chronic pain, chronic fatigue syndrome, sleep disorders, PTSD, migraines, and cognitive decline. ²¹

As evidenced by the retrieval of only a single completed study ⁴¹ and a single RCT protocol ⁴⁰ investigating effectiveness of neurofeedback for cancer survivors experiencing cancer fatigue or cognitive impairment, the results of this review demonstrate that the field of oncology has not yet evaluated this noninvasive therapy, which has the potential to ameliorate symptoms in cancer patients and cancer survivors. All the included studies related to fatigue and cognition were published since 2000, suggesting this area of research is relatively new. However, each of these studies reported positive results for at least 1 fatigue or cognitive outcome in a variety of clinical (traumatic brain injury, fibromyalgia, CNS problems) and nonclinical (college students, adults, elderly) populations. Despite relatively small sample sizes, statistically significant results were obtained. This suggests that neurofeedback addresses an underlying mechanism for fatigue and cognition common to these different issues or is responsive to a variety of neuropsychological pathologies either through targeted or generalized approaches.

Results of this review are consistent with reports in the literature that describe neurofeedback as safe and well tolerated. Although only 4 studies commented on side effects, none of the studies reported subjects dropping out due to side effects. Neurotherapy rarely produces negative side effects, which is a major reason for its use. ²⁸ Adverse effects reported in the literature appear to be associated with volitional types of neurofeedback protocols that attempt to push the brain in specific directions, which can occasionally result in side effects and when improperly done, exacerbation of symptoms. ⁶⁴ In the case of nonvolitional neurotherapy, such as NLD approaches, the practitioner does not push the brain in any particular direction but cues the brain to strengthen its own self-regulatory networks. ³² Reported side effects seem to be exacerbation of existing symptoms or complaints, and are usually very transient, dropping away after only a few sessions. These results are consistent with findings of a survey of providers using an NLD system. ⁶⁵

Of the 3 studies included in this review that reported side effects or adverse effects, 1 study involved peak alpha frequency training ⁴⁷ and 2 involved LENS therapy. The peak alpha training pushed the brain in a specific direction, which improved some cognitive measures at the expense of visual memory. ⁴⁷ The LENS neurofeedback interventions, which are nonvolitional, resulted in transient exacerbation of symptoms such as fatigue and pain. Participants in the single NLD neurofeedback study did not report any side effects.

There appears to be no single conceptual model to describe the process and outcomes of neurofeedback among practitioners or manufacturers.^26,27 In this review, a variety of neurofeedback systems, approaches and protocols were employed for the purpose of ameliorating symptoms of chronic fatigue and improving cognition. All of the included studies reported positive results for at least one fatigue or cognitive outcome suggesting that multiple approaches may be effective.

The lone study that evaluated neurofeedback with breast cancer survivors, although not an RCT, demonstrated promising results using an NLD approach: 23 women demonstrated significant improvements on all 4 cognitive measures and the fatigue scale after only 20 sessions with no side effects, adverse effects or dropouts reported. These results suggest that the NLD approach warrants further study through randomized controlled trials and larger sample sizes to confirm efficacy and safety as well as to establish the optimum range or number of sessions required to achieve lasting results. Overall, results suggest that protocols for any of the neurofeedback approaches should include a minimum of 10 to 20 sessions to demonstrate effectiveness, recognizing that individuals may require fewer or more sessions to alleviate their symptoms.

This systematic review was conducted using JBI methodology, which includes studies using different research designs. Different appraisal tools are used to evaluate studies that are experimental (randomized control/pseudo-randomized trial) versus observational/descriptive studies. Although observational/descriptive studies cannot be used to determine causation, they suggest relationships that could be further investigated through RCTs if appropriate, and thus were in included in this review, which explores a novel intervention that has not been well researched.

Collectively, results of the studies included in this review represent at best level 2 evidence due to the inclusion of several small RCTs and quasi-experimental studies according to the JBI levels of evidence in which level 1 evidence, the strongest evidence would consist of results of a meta-analysis (with homogeneity) of experimental studies or one or more large experimental studies with narrow confidence intervals. ⁴⁴

Limitations

The greatest limitation of this review is the small sample sizes used in both the experimental and descriptive studies. Confounding factors such as concurrent medication use were often not described. Methodological quality of the majority of the studies was only moderate, with few blinded, RCTs included. Use of different outcome measurement tools limits the ability to make comparisons between different neurofeedback protocols and approaches. As well the diversity of neurofeedback approaches and systems makes it difficult to categorize and explain the overall positive results.

Conclusions

Despite issues with methodological quality, the overall positive findings and few reported side effects associated with neurofeedback for fatigue and cognitive impairment suggest this innovative CAM therapy could be helpful in alleviating these symptoms. Currently, there is insufficient evidence that neurofeedback is an effective therapy for management of fatigue and cognitive impairment in cancer survivors; however, these promising results support the need for further research with this patient population. More information about which neurofeedback technologies, approaches and protocols could be successfully used with cancer survivors and with minimal side effects is needed. Clinical trials comparing the effectiveness of specific neurofeedback protocols and approaches to sham neurofeedback and/or to other neurofeedback protocols are urgently needed to move this field forward. Trials should also compare neurofeedback with other approaches that encourage healthy self-regulation of the brain such as physical activity and mindfulness-based stress reduction strategies. Study limitations that should be addressed in future research include use of randomization and double-blinding, larger sample sizes and increased statistical power to increase the quality of individual studies as well as the overall level of evidence to efficacy and safety of various neurofeedback approaches.

This review will provide preliminary data on which to build a protocol for a pilot RCT in a sample of cancer survivors to determine feasibility, recruitment strategies, validate instruments, and establish effect size for calculation of sample size for a larger clinical trial to evaluate the effectiveness and safety of neurofeedback in this population. In the long term, the proposed program of research may help cancer survivors with long term symptoms for whom other approaches have failed, or who do not want to continue to use medications, as well as provide scientific evidence of efficacy. The results will have significance to nurses and physicians in oncology and primary care settings who provide follow-up care and counseling to cancer survivors experiencing debilitating symptoms such as cognitive impairment (chemobrain) and cancer fatigue.