REVIEW

Neurofeedback for Attention-Deficit/Hyperactivity

Disorder: Meta-Analysis of Clinical and
Neuropsychological Outcomes From Randomized
Controlled Trials
Samuele Cortese, MD, PhD, Maite Ferrin, MD, PhD, Daniel Brandeis, PhD, Martin Holtmann, MD,
Pascal Aggensteiner, MA, David Daley, PhD, Paramala Santosh, MD, PhD, Emily Simonoff, MD, PhD,
Jim Stevenson, PhD, Argyris Stringaris, MD, PhD, MRCPsych,
Edmund J.S. Sonuga-Barke, PhD, on behalf of the European ADHD Guidelines Group (EAGG)

Objective: We performed meta-analyses of randomized only frequency band training trials, the most common
controlled trials to examine the effects of neurofeedback neurofeedback approach, were analyzed separately.
on attention-deficit/hyperactivity disorder (ADHD) Effects on laboratory measures of inhibition (SMD ¼ 0.30,
symptoms and neuropsychological deficits in children and 95% CI ¼ 0.10 to 0.70) and attention (SMD ¼ 0.13, 95%
adolescents with ADHD. CI ¼ 0.09 to 0.36) were not significant. Only 4 studies
directly assessed whether learning occurred after neuro-
Method: We searched PubMed, Ovid, Web of Science, feedback training. The risk of bias was unclear for many
ERIC, and CINAHAL through August 30, 2015. Random- Cochrane Risk of Bias domains in most studies.
effects models were employed. Studies were evaluated
with the Cochrane Risk of Bias tool.
Conclusion: Evidence from well-controlled trials with
Results: We included 13 trials (520 participants with probably blinded outcomes currently fails to support
ADHD). Significant effects were found on ADHD symp- neurofeedback as an effective treatment for ADHD.
toms rated by assessors most proximal to the treatment Future efforts should focus on implementing standard
setting, that is, the least blinded outcome measure (stan- neurofeedback protocols, ensuring learning, and opti-
dardized mean difference [SMD]: ADHD total mizing clinically relevant transfer.
symptoms ¼ 0.35, 95% CI ¼ 0.110.59; inattention ¼ 0.36,
95% CI ¼ 0.090.63; hyperactivity/impulsivity ¼ 0.26, Key words: ADHD, neurofeedback, nonpharmacological
95% CI ¼ 0.080.43). Effects were not significant when treatment, meta-analysis, risk of bias
probably blinded ratings were the outcome or in trials
with active/sham controls. Results were similar when J Am Acad Child Adolesc Psychiatry 2016;55(6):444–455.

A ttention-deficit/hyperactivity disorder (ADHD) is a

common neurodevelopmental disorder characterized
by age-inappropriate and impairing inattention and/
or hyperactivity/impulsivity.1,2 Among currently available
treatment options, psychostimulant and nonstimulant medi-
electroencephalogram (EEG) indices of interest are converted
into visual or acoustic signals and fed back automatically in real
time to the patient. For instance, cortical activity may be rep-
resented by the height or speed of a ball, plane, or cartoon
character presented using animation on a computer screen. In
cations are efficacious, at least in the short term, and widely this case, learning occurs when the object rises, falls, or ad-
used.3 Nonpharmacological interventions—both dietary and vances more quickly in response to patients’ regulated changes
psychological—have also been extensively investigated.4-7 in brain activity. Two general neurofeedback approaches have
Among nonpharmacological approaches, neurofeedback has been used to treat ADHD: frequency band training (FBT) and
been considered a promising ADHD treatment strategy since slow cortical potential training (SCP). When applied to ADHD,
the early 1970s.8-10 When applied to ADHD, neurofeedback is the former is intended to target alterations in cortical electrical
intended to reduce ADHD symptoms by targeting aberrant oscillations thought to be associated with ADHD, namely ele-
patterns of brain activity thought to underpin the condition. vations of slow, relative to fast, brainwave activity, especially in
Neurofeedback is implemented through the training of self- the frontal lobes (e.g., theta versus beta frequency11). The latter
regulation using operant reinforcement procedures; learning aims to regulate cortical excitation thresholds by focusing on
of self-regulation is thus a key mechanism. To achieve this aim, activity generated by external cues (similar to event-related
potentials), focusing primarily on EEG components registered
in the late latency range, that is, several seconds after the cue.
For instance, this form of training has been used to target the
Supplemental material cited in this article is available online. contingent negative variation (CNV) that occurs during this
time window and is involved in effective preparation, decision

JOURNAL OF THE AMERICAN ACADEMY OF C HILD & ADOLESCENT PSYCHIATRY

444 www.jaacap.org VOLUME 55 NUMBER 6 JUNE 2016
 META-ANALYSIS OF NEUROFEEDBACK OUTCOMES IN ADHD

making, and time estimation, which have all been found to be METHOD
deficient in individuals with ADHD, or at least in subgroups of The EAGG protocol was originally registered on the Inter-
them.12,13 national Prospective Register of Systematic Reviews (PROSPERO;
The efficacy of nonpharmacological treatments for http://www.crd.york.ac.uk/PROSPERO, protocol number:
ADHD, including neurofeedback, has been subject to a CRD42011001393). As in previous work,4,5 the original protocol was
number of earlier meta-analytic reviews.14-16 However, these adapted to take account of the broader scope of this systematic
have sometimes been difficult to interpret because of the review/meta-analysis. Most crucially, given that the scope of this
analysis included neuropsychological measures, the mandatory
inclusion of studies with weak experimental designs (e.g., no
requirement for studies to have ADHD symptoms-related outcomes
control arm, nonrandom allocation, or the use of nonblinded
no longer applied (i.e., we included also studies presenting only
measures), as discussed by Sonuga-Barke et al.17 On behalf of neuropsychological outcomes).
the European ADHD Guidelines Group (EAGG), Sonuga-
Barke et al.17 attempted to address these limitations
through a meta-analysis of nonpharmacological in- Inclusion and Exclusion Criteria
terventions that included only randomized controlled trials To ensure high levels of methodological adequacy as recommended
(RCTs). It also addressed the issue of blinding by comparing by the Cochrane group and to avoid the inevitable bias caused by
outcomes rated by individuals judged to be most proximal dependence on investigators agreeing to provide data from unpub-
to the therapeutic setting (often parents poorly blinded and lished studies,20 only published studies were included. Only RCTs
invested in the therapeutic outcome) and those provided by using neurofeedback training were retained. Participants in the trials
reporters judged to be probably blinded. They found that the were required to be between 3 and 18 years of age and to have a
effects of neurofeedback on ADHD total symptoms based on diagnosis of ADHD (any subtype) or hyperkinetic disorder (HKD) or
to meet accepted cut-offs on validated ADHD symptom rating scales.
most proximal ratings were highly significant (standardized
Trials that selected children with ADHD who had rare comorbid
mean difference [SMD] ¼ 0.59, 95% CI ¼ 0.31, 0.87). How-
disorders (e.g., Fragile X syndrome) were excluded. Control condi-
ever, when only probably blinded measures were used, the tions allowed were “treatment as usual,” “wait list,” “active,” or
effects became nonsignificant (SMD ¼ 0.29; 95% CI ¼ 0.02 “placebo/sham” (i.e., involving other forms of alternative training
to 0.61). More recently, Micolaud-Franchi et al.18 followed a regimen). As per the EAGG protocol, trials in which neurofeedback
similar approach, focusing their analyses on ADHD core was compared only with optimized medication or in which neuro-
symptoms, but with a smaller set of studies (n ¼ 5) limited to feedback was added to optimized medication were excluded. Trials in
trials with particular control conditions. As in Sonuga-Barke which medication was part of background normal clinical provision
et al.,17 they found a significant, positive effect of neuro- in either the control or the active arm were included.
feedback on ADHD core symptoms when considering most
proximal raters. Probably blinded scores were attenuated
and were significant only for symptoms of inattention. Search Strategy
Applying the same meta-analyses protocol used in recent Details about the search strategy/syntax for each database are re-
ported in Supplement 1, available online. The final search was
EAGG reviews of behavioral interventions5 and cognitive
updated on August 30, 2015. Independent searches were conducted
training,4 we here extend the focus of meta-analytic evidence
by 2 authors (S.C. and M.F.), leading to the same number of references.
relating to neurofeedback for ADHD in a number of ways.
First, we included, among the outcomes, not only specific
ADHD behavioral dimensions (i.e., inattention and impul- Outcome Measures
sivity/hyperactivity) but also ADHD-related neuropsycho- To provide analytical robustness and in line with previous EAGG
logical deficits such as inhibitory dysfunction. The latter may meta-analyses,4,5,17,20 analyses of outcome domains were considered
be important, as they may take us closer to neural mediators reliable only if at least 5 RCTs were available. The planned outcomes
of the behavioral effects of neurofeedback.9 Second, we included the following: ADHD symptoms (total ADHD and inat-
addressed the relative efficacy of different types of neuro- tention and hyperactivity/impulsivity symptoms separately), neu-
feedback by restricting subanalyses to specific types of ropsychological laboratory-based measures, measures of academic
treatment protocols, namely, FBT. Third, we examined the functioning, and rating of severity of symptoms of comorbid con-
impact of different aspects of trial design (e.g., use of a ditions (e.g., oppositional defiant disorder or anxiety disorders).
sham/placebo design) or pragmatic “dosage” characteristics
of neurofeedback implementation (i.e., number of sessions).
Fourth, we addressed the crucial question of whether Study Selection
neurofeedback-related learning at the neural level was Retrieved references were independently screened and blindly
double coded for eligibility by 2 authors (S.C. and M.F.). Any
investigated and/or demonstrated in available trials.9 Fifth,
disagreement was resolved by a senior author (E.S.-B.).
we examined whether the neurofeedback protocols used in
these studies could be considered “standard” in terms of the
criteria discussed by Arns et al.,19 which include elements Study Bias Assessment
related to EEG bands/measures, electrode placement and Study quality was assessed independently by pairs of raters from
type, and feedback following learning. Finally, we applied, the authorship group using the Cochrane RoB tool.20 The RoB do-
for the first time in a meta-analysis of neurofeedback for mains included selection bias, performance bias, detection bias,
ADHD, a rigorous assessment of study bias, namely, the attrition bias, and other bias. Any disagreement was resolved
Cochrane Risk of Bias tool (RoB).20 through consensus.

JOURNAL OF THE AMERICAN ACADEMY OF C HILD & ADOLESCENT PSYCHIATRY

VOLUME 55 NUMBER 6 JUNE 2016 www.jaacap.org 445
CORTESE et al.

Data Extraction and Statistical Analysis considering only trials with an active/sham control, the ef-
Trial information was entered into RevMan 5.3.21 Data extraction fects dropped to nonstatistically significant levels for total
was independently performed and cross-checked by the first 2 au- ADHD and inattention symptoms. The effect was significant
thors. SMD was calculated as mean pre- to posttreatment change in for hyperactivity/impulsivity but with a small effect size
the intervention group minus the mean pre- to posttreatment change (SMD ¼ 0.25), and the 95% CI was wide (CI ¼ 0.030.47).
in the control group, divided by the pooled pretest standard devi- When probably blinded outcomes were analyzed, effect
ation with a bias adjustment.22 SMDs for each trial were combined sizes for ADHD outcomes dropped further, and none were
using the inverse variance method. Given the inherent heterogeneity
significant (Figure 3). Results were also not significant when
of studies, random effects models were used. The I2 statistic was
calculated to estimate between-trial SMD heterogeneity. When
considering only probably blinded measures from trials with
multiple measurements were available for an outcome, the most active/sham control (Table 2). When considering only trials
frequently reported outcome across trials or the outcome that was evaluating FBT, results were significant for ADHD total
judged to better tap the core of the construct was selected. To be (SMD ¼ 0.37, 95% CI ¼ 0.090.64) and hyperactive/
consistent with the EAGG protocol, we considered as probably impulsive symptoms (SMD ¼ 0.26, 95% CI ¼ 0.060.46)
blinded those outcomes rated by an individual who was likely un- from most proximal raters, but not for ADHD inattentive
aware of treatment allocation. Most proximal ratings were based on symptoms or any ADHD symptoms rated by probably
assessors close to the therapeutic setting and often not blinded. blinded assessors. There were insufficient trials (n ¼ 230,32)
These ratings typically constituted a trial’s own primary outcome for an analysis focused on SCP training (2 studies also used
measure and were therefore the assessment most available for
both SCP and FBT31,33). When pooling only trials with no/
analysis.
Selection of most proximal and probably blinded ratings was
low medication, results were significant only for ADHD
based on independent judgments and consensus of 3 authors inattentive symptoms, most proximal (SMD ¼ 0.59, 95%
(S.C., M.F., and E. S.-B.). Where 2 or more probably blinded ratings CI ¼ 0.310.88) and hyperactive/impulsive symptoms,
were available (which was sometimes the case on sham-controlled most proximal (SMD ¼ 0.32, 95% CI ¼ 0.060.58).
trials), what was deemed the best probably blinded outcome When the analysis was restricted to only the 7 trials that
was selected for analysis, as in Cortese et al.4 When there were 2 or met Arns et al.19 criteria for use of standard protocol, the
more neuropsychological outcome measures, the one most SMDs increased for most proximal outcomes, although the
frequently reported across the relevant trials was selected. Four 95% CIs were wide (Table 2). There were only 3 trials27,30,31
sensitivity analyses were conducted. The first included only trials that both used a standard protocol and had probably blin-
using an active or sham control. The second was restricted to
ded measures. In an exploratory analysis with these 3 trials,
FBT trials. The third examined the effect of co-treatment with
medication and was restricted to studies with no/low levels of
the effect on total ADHD symptoms was significant (SMD ¼
medication (<30% of participants on medication, as per EAGG 0.36), although the 95% CI was large (0.040.69).
protocol). The final sensitivity analysis included only studies The analysis of neuropsychological outcomes required
meeting the criteria defining a standard neurofeedback method- the pooling of diverse neuropsychological measures within
ology as described by Arns et al.19 (see Supplement 2, available general domains (Table 1). Effects on laboratory measures of
online). Meta-regression was conducted to assess the effects of inhibition (SMD ¼ 0.30, 95% CI ¼ 0.10 to 0.70) and
number of training sessions. Publication bias was assessed with attention (SMD ¼ 0.13, 95% CI ¼ 0.09 to 0.36) were not
funnel plots and Egger tests. Analyses were conducted using significant.
RevMan 5.321 and STATA 13.1.23 Direct evidence that neurofeedback training led to
learning was gathered in only 4 trials27,29,33,36 and was
positive for only 1 trial.27 In 1 study,29 there was a partial
RESULTS learning effect. In Heinrich et al.,32 there was indirect evi-
A total of 13 trials24-36 met entry criteria. Figure 1 presents dence of learning: the increase in the contingent negative
the Preferred Reporting Items for Systematic Reviews and variation (a slow cortical potential) suggests that children
Meta-Analyses (PRISMA) selection flowchart. Table S1, learned what was trained.
available online, provides a list of excluded papers with Studies varied considerably in terms of risk of bias
reasons for exclusion. Retained studies included a total of (Figures S1 and S2, available online). For approximately half
520 participants with ADHD. Table 1 gives information of the RoB domains across studies, the level of risk was
about the characteristics of the retained trials. Table 2 unclear. For those categories that could be determined, 8
summarizes results of all analyses. There were fewer than trials were rated as having a high risk of bias in at least 1
5 studies available to examine some of the planned out- domain. In general, the major concern (i.e., high risk of bias)
comes such as working memory, parent ratings of execu- related to blinding of participants, personnel, and assessors.
tive functioning (e.g., Behaviour Rating Inventory of Among “other bias,” funding was not clearly stated in a
Executive Function37), academic functioning (e.g., reading sizable portion of studies.
or arithmetic ability), or comorbid conditions (e.g., ODD, Funnel plots and Egger test results (Figures S3 and S4,
anxiety). Therefore, we did not perform analyses for such available online) suggested little evidence of publication
outcomes. bias, although the number of trials was insufficient to
When most proximal assessments were the outcome, establish a reliable estimate. Finally, meta-regression ana-
there was a small-to-moderate (SMD < 0.5) but significant lyses did not support a relationship between the number of
effect on inattention, impulsivity/hyperactivity, and total training sessions and most proximal or probably blinded
ADHD symptoms (Figure 2). In sensitivity analyses outcomes (Figures S5 and S6, available online).

JOURNAL OF THE AMERICAN ACADEMY OF C HILD & ADOLESCENT PSYCHIATRY

446 www.jaacap.org VOLUME 55 NUMBER 6 JUNE 2016
 META-ANALYSIS OF NEUROFEEDBACK OUTCOMES IN ADHD

FIGURE 1 Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram of selection of studies (last
search updated on August 30, 2015). Note: Reasons for exclusion of each article are reported in Table S1. aFour articles included in
the search results (29,35,36,46) refer to the same sample as refered to by another four articles (44,43,45,31, respectively), so the PRISMA
flowchart does not count them twice.
Identification

Records identified through Additional records identified

database search through other sources
(n = 264) (n = 4)

Records after duplicates removed

(n = 247)
Screening

Records screened Records excluded

(n = 247) (n = 204)

Full-text articles assessed Full-text articles

Eligibility

for eligibility excluded, with reasons

(n = 43) (n = 26)a

Studies included in
qualitative synthesis
(n = 13 + 4)a
Included

Studies included in
quantitative synthesis
(meta-analysis)
(n = 13 + 4)a

DISCUSSION neurofeedback procedures. Although it does not provide a

The rationale for the use of neurofeedback for ADHD rests definitive statement as to the value of neurofeedback, the
on the idea that promoting normalization or self-regulation current analysis has clarified a number of issues.
of brain activity will translate into improved cognitive and In general, the effect size estimates in the current analysis
behavioral control that is deficient in individuals with are substantially smaller compared to the previous one by
ADHD. A previous meta-analysis17 based on a limited Sonuga-Barke et al.17 This is due to the smaller effect sizes
number of trials (n ¼ 8) was inconclusive with regard to the reported in the most recent trials not included in Sonuga-
efficacy of neurofeedback for ADHD symptoms. Although Barke et al.17 For instance, SMDs dropped for most prox-
there were moderate but highly significant effects (SMD ¼ imal and probably blinded total ADHD outcomes by 41%
0.59; 95% CI ¼ 0.310.87) for ADHD outcomes rated by and 49%, respectively, although the analysis based on most
most proximal assessors, these were not significant when proximal ratings of ADHD core symptoms remained sig-
only probably blinded outcomes were considered (SMD ¼ nificant. Crucially, as in the previous meta-analysis,17 when
0.29; 95% CI ¼ 0.02 to 0.61). The current meta-analysis, the risk of biased effect size estimates was reduced either by
including an additional 5 RCTs, provides what is likely to selecting probably blinded outcomes or by limiting analysis
be a more reliable estimate of the effects of neurofeedback to trials with a high-quality control arm (sham or active),
for ADHD. Importantly, the additional statistical power also effects were no longer significant. Importantly, and in
allowed us to explore the impact of neurofeedback type, the contrast to previous meta-analyses,15,18 there was no evi-
effect of type of control, the impact of number of neuro- dence for a particular benefit with regard to inattention
feedback sessions, and the value of using standard symptoms, which have previously been hypothesized to be

JOURNAL OF THE AMERICAN ACADEMY OF C HILD & ADOLESCENT PSYCHIATRY

VOLUME 55 NUMBER 6 JUNE 2016 www.jaacap.org 447
448

CORTESE et al.
TABLE 1 Characteristics of Studies Included in Meta-Analysis
www.jaacap.org

Design Training Sample Outcomes

c
N Meds (%) ADHD Symptoms Neuropsychological Outcomes
Standard Sessions, Follow- t t
Triala Control Type NF Protocolb n Up (mo) c c Age (mo) M-PROX P-BLIND Inhibitory Control Attention
26 e
Arnold Sham FTB no 40 2 25 0 72-144 Parent Teacher BRC
placebod 11 0
Bakhshayesh27 EMG FTB yes 30 6f 18 22 72-168 Parent Teacher CPT (commission) CPT (omission)
biofeedback 17 18
Beauregard28 No treatment FTBg yes 40 N/S 15 0 96-144 Parent Counting Stroop CPT (integrated
5 0 (interference) visual and
auditory)
Bink29 TAU FTBh no 37 12 45 44.4 193.2  39.6 (T) Parent Color Stroop D2 attention (total
26 61.5 194.4  40.8 (C) (Interference) correct)
23i,j
Christiansen30 Self-manag. SCP yes 30 12 58 (tot.) 101.04  16.08 Parent Teacher
Gevensleben31 AT SCP þ FTBk yes 36 6 59 8.5 118  15 (T) Parent Teacher ANT, Orienting ANT,
35 2.9 112  14 (C) (data from46, that Conflict
refers to the same (data from46,
JOURNAL

sample of31) that refers to the

same
OF THE

sample of31)
Heinrich32 WL SCPl yes 25 N/S 13 46.1 90-165 Parent CPT (commission) CPT (omission)
AMERICAN ACADEMY OF C HILD & ADOLESCENT PSYCHIATRY

9 44.4
Holtmann25 ATm FTBl yes 20 N/S 20 79.4n 123.6  Parent Stop Signal Task
14 14.4 (mean)
Linden24 WL FTB yes 40 N/S 8 0 110 (mean) Parent
VOLUME 55 NUMBER 6 JUNE 2016

6 0
Maurizio33 EMG SCP þ FTBo no 36 N/S 13 7.6i 102-154.8 Parent Teacher D2 attention
biofeedback 12 8.3i (total score)
Steiner34p ATm, WLq FTBh no 23 N/S 9 60 148.8  10.8 Parent Teacher
11
Steiner35 CT, WLq FTBh no 40 6r 34 44.1 100.8  13.2 (T) Parent BOSS
36 55.5 100.8 
13.2 (C)
 TABLE 1 Continued
VOLUME 55 NUMBER 6 JUNE 2016
JOURNAL OF THE AMERICAN ACADEMY OF C HILD & ADOLESCENT PSYCHIATRY

Design Training Sample Outcomes

c
N Meds (%) ADHD Symptoms Neuropsychological Outcomes
Standard Sessions, Follow- t t
Triala Control Type NF Protocolb n Up (mo) c c Age (mo) M-PROX P-BLIND Inhibitory Control Attention
45 t u
VanDongen Placebo NF FTB no 30 6 22 54.5 126.0  26.4 (T) Investigator Teacher
Vollebregt36s 19 73.7 128.4 
27.6 (C)
Note: Studies are listed in alphabetical order. ADHD ¼ attention-deficit/hyperactivity disorder; ANT ¼ Attention Network Test; AT ¼ attention training; BOSS ¼ Behavioral Observation of Students in Schools; BRC ¼ brain
resource center computer-based normed neuropsychological test; C ¼ control group; CPT ¼ continuous performance test; CT ¼ cognitive training; EMG ¼ electromyography; FTB ¼ frequency theta/beta; NF ¼
neurofeedback; N/S ¼ not specified; SCP ¼ slow cortical potential; SMR ¼ sensory motor rhythm; SST ¼ Stop Signal Test; T ¼ treatment condition; TAU ¼ treatment as usual; WL ¼ waiting list.
a
In alphabetical order, followed by study reference number.
b
Standard methodology described elsewhere.19
c
N is the number of individuals in the treatment (T) and control (C) conditions.
d
Equal intensity and duration.
e
EEG recorded from an electrode placed at position Cz vs. the ears as reference
f
Results of the follow-up were not published in this article.
g
SMRþbeta.
h
SMR.
i
Children were off medication 48 hours prior to all the assessments.
j
Number in each group not specified.
k
SMRþbeta Cz-ears.
l
Calculated from Cz vs mastoids.
m
Standard computer format.
n
A total of n¼27 were medicated (no indication of how many in each group).

META-ANALYSIS OF NEUROFEEDBACK OUTCOMES IN ADHD

o
B¼beta, LORETA Tomography.
p
We used parent #1 because parent #2 measures were available for a smaller sample (n ¼ 9 for parent #1 and n ¼ 5 for parent #2 in NF and n ¼ 11 for parent #1 and n ¼ 9 for parent #2 in the waiting list group);
a discussion of the outcomes selected from this study is available elsewhere.47,48
q
The WL arm was used as comparator in the analyses.
r
The results of the follow-up phase are published elsewhere.43
s
These 2 articles refer to the same study and present analyses on different outcomes.
t
Individualized, mainly T/B, B ¼ SMR.
u
Data reported only at endpoint (15 wk).
www.jaacap.org
449
CORTESE et al.

TABLE 2 Summary of Results

Effect of Intervention Heterogeneity
2
Outcome Trials Included Measure Trials, n SMD 95% CI p I p
ADHD
symptoms
Total All MPROX 13 0.35 0.11, 0.59 .004 41 .06
PBLIND 8 0.15 0.08, 0.38 .20 0 .74
Active/sham MPROX 7 0.22 0.08, 0.52 .14 35 .16
PBLIND 6 0.20 0.05, 0.45 .12 0 .66
FBT MPROX 9 0.37 0.09, 0.64 .01 36 .13
PBLIND 5 0.03 0.29, 0.35 .84 0 .76
Low medication MPROX 7 0.39 0.01, 0.79 .05 48 .08
PBLIND 5 0.26 0.01, 0.54 .06 0 .76
Standard protocol MPROX 7 0.45 0.02, 0.88 .04 53 .05
Inattention All MPROX 11 0.36 0.09, 0.63 .009 43 .07
PBLIND 7 0.06 0.24, 0.36 .70 41 .12
Active/sham MPROX 6 0.26 0.10, 0.63 .16 49 .08
PBLIND 5 0.21 0.09, 0.50 .17 16 .31
FBT MPROX 9 0.33 0.00, 0.67 .05 51 .04
PBLIND 5 0.04 0.37, 0.28 .79 19 .29
Low medication MPROX 6 0.59 0.31, 0.88 .0001 0 .63
Standard protocol MPROX 5 0.55 0.01, 1.09 .05 58 .05
Hyper/imp All MPROX 10 0.26 0.08, 0.43 .004 0 .80
PBLIND 7 0.17 0.05, 0.39 .13 0 .59
Active/sham MPROX 6 0.25 0.03, 0.47 .03 0 .92
PBLIND 5 0.15 0.11, 0.41 .26 0 .80
FBT MPROX 8 0.26 0.06, 0.46 .01 0 .62
PBLIND 5 0.15 0.15, 0.44 .33 6 .37
Low medication MPROX 5 0.32 0.06, 0.58 .02 0 .44
Neuropsychological
test performance
Attention All Objective 8 0.13 0.09, 0.36 .26 0 .72
FBT Objective 5 0.11 0.16, 0.38 .43 0 .76
Inhibition All Objective 6 0.30 0.10, 0.70 .15 56 .05
Note: Pooled standardized mean differences (SMD) are shown for each outcome. Positive SMDs indicate that neurofeedback is more efficacious than control condition.
Significant effects are indicated in boldface type. There were insufficient (n < 5) trials for the following: standard neurofeedback (NF), probably blinded rater (PBLIND)
total (tot; n ¼ 3), PBLIND inattention (In; n ¼ 3); standard protocol, most proximal rater (MPROX), hyperactivity/impulsivity (hyp/imp; n ¼ 4), and PBLIND (tot, In,
hyper/imp) (n ¼ 3); trials using a version of frequency band training including thetaebeta ratio training (FBT), inhibition (n ¼ 4 trials); low medication, inattention,
PBLIND (n ¼ 4) and hyp/imp, PBLIND (n ¼ 4). Active/sham ¼ only trials with an active/sham control arm; ADHD ¼ attention-deficit/hyperactivity disorder; All ¼ all
trials meeting inclusion criteria with available measures; SMT ¼ self-management therapy; TAU ¼ treatment as usual.

more amenable to neurofeedback. This was also reflected in improvements in the setting in which the treatment was
our failure to find effects on laboratory measures of attention delivered, but these effects failed to generalize to more distal
such as the continuous performance test. settings in which blinded measures were recorded. This
Previously, Sonuga-Barke et al.17 have argued that the explanation seems unlikely, as probably blinded measures
substantial drop in SMDs between most proximal and were also collected in the treatment setting for some trials,
probably blinded analyses is likely to be the result of biases and these followed the same pattern. The type of neuro-
in perception in favor of the active treatment when one relies feedback protocol implemented in the trials did not seem to
on observations by raters aware of treatment allocation. be an important factor in determining the results. When we
However, there are some other explanations. For instance, it restricted the analysis to trials using some form of FBT such
is possible that probably blinded ratings are, for some as that focusing on alteration of the ratio between slow theta
reason, less sensitive to change than most proximal mea- and faster beta oscillations, there was no increase in the
sures, perhaps either because of the instruments used or the effects of treatment. There was an insufficient number of
person rating. Teachers, for instance, may be less sensitive to trials (n ¼ 2)30,32 using SCP training exclusively to state
change than parents. However, essentially the same ques- any firm assertion about this approach. The results
tionnaires were completed by both of these types of raters. It raise the question of whether current training protocols
is also possible that proximal ratings accurately reflected real have the appropriate treatment target. For instance, the

JOURNAL OF THE AMERICAN ACADEMY OF C HILD & ADOLESCENT PSYCHIATRY

450 www.jaacap.org VOLUME 55 NUMBER 6 JUNE 2016
 META-ANALYSIS OF NEUROFEEDBACK OUTCOMES IN ADHD

FIGURE 2 Forest plots for meta-analysis of effects on attention-deficit/hyperactivity disorder (ADHD) core symptoms assessed
by most proximal (MPROX) raters. Note: A) ADHD total symptoms; B) inattentive symptoms; C) hyperactive/impulsive symptoms.
SE ¼ standard error; Std ¼ standardized.

developmental stability of EEG frequency band alterations studies rather than the weakness of neurofeedback as such.
in ADHD from childhood into adulthood has been ques- Indeed, the current set of 13 trials, taken as a whole, had a
tioned,11,38,39 although the most commonly observed ADHD number of methodological short-comings. First, only 4
effects with regard to evoked brain responses relate to early studies tested,27,29,33,36 and then only 1 study,27 reported
rather than later components not targeted in current neu- directly and positively whether neurofeedback training had
rofeedback protocols. Therefore, the rationale for theta–beta actually led to learning as indexed by changes/improve-
feedback has been very critically discussed.9,40 ment at the electrophysiological level. In the 1 positive
The value of a treatment meta-analysis is, of course, trial,27 the mediating role of changes in the electrophysio-
constrained by the methodology of the trials that it includes. logical signature leading to changes at the symptom level
It is therefore possible that the results of our meta-analysis was not investigated. This is a crucial point, because if
reflect the methodological weaknesses of the included neurofeedback cannot bring about the expected changes at

JOURNAL OF THE AMERICAN ACADEMY OF C HILD & ADOLESCENT PSYCHIATRY

VOLUME 55 NUMBER 6 JUNE 2016 www.jaacap.org 451
CORTESE et al.

FIGURE 3 Forest plots for meta-analysis of effects on attention-deficit/hyperactivity disorder (ADHD) core symptoms assessed by
probably blinded (PBLIND) raters. Note: A) ADHD total symptoms; B) inattentive symptoms; C) hyperactive/impulsive symptoms.
SE ¼ standard error; Std ¼ standardized.

the neural level, then treatment effects are more likely to be Interestingly, only 54% of the studies in the current analysis
artifacts of some other, nonspecific aspect of the training. met such a threshold. When we restricted the analysis to that
Such a situation may be similar to cognitive training tar- subset of trials, our results were somewhat mixed; the effect
geting but not improving working memory, or to a drug size for total ADHD and inattention increased by about 20%.
with an established neurotransmitter profile not reaching the However, for all most proximal outcomes, 95% CIs were
corresponding neural target system in the patient’s brain. On wide and close to nonsignificance. Unfortunately, there were
the other hand, tests for learning of neural self-regulation, only 3 studies with a standard protocol including probably
and addressing relations between learning and clinical blinded ratings,27,30,31 so that firm conclusions on the value
improvement, would also need to consider more complex of standardized protocols as defined by Arns et al.19 cannot
models and alternative mechanisms, for example allowing be drawn. In addition, the level of methodological rigor
for initial, delayed, and nonlinear types of learning and specifically related to RCT conduct, as explored by the RoB
translation,41 before concluding that effects are nonspecific. tool, was in general unclear. The level of blinding was un-
Further research should address whether possible neuronal clear or insufficient in many studies. In addition, 1 particu-
modifications underpin putative behavioral changes in larly striking omission in the majority of trials was the report
ADHD symptoms following neurofeedback. of possible potential conflicts of interest. This would seem to
A number of groups have defined what constitutes a be a major oversight in the current literature, given the
standard neurofeedback protocol in terms of the number of growing presence of neurofeedback training companies in
training sessions, the reinforcement parameters operating, the commercial treatment marketplace. Given these limita-
the EEG montage, etc. (see Vernon et al.42 for a justification). tions, it seems that without evidence for the learning of self-

JOURNAL OF THE AMERICAN ACADEMY OF C HILD & ADOLESCENT PSYCHIATRY

452 www.jaacap.org VOLUME 55 NUMBER 6 JUNE 2016
 META-ANALYSIS OF NEUROFEEDBACK OUTCOMES IN ADHD

regulation, and given the widespread use of nonstandard different neuroimaging approaches with neurofeedback
neurofeedback protocols, it is difficult to draw definitive training might be a useful approach in the future.
conclusions about the ultimate value of neurofeedback Finally, the range of number of sessions across studies
approaches. may have been too restricted to allow the detection of
The EAGG has also recently completed meta-analyses possible effect of session number.
for behavioral interventions and cognitive training using In summary, the current meta-analysis shows that evi-
the same core protocol as used here.4,5 There are striking dence from well-controlled trials with probably blinded
similarities but also some differences between the results of outcomes does not support neurofeedback as an effective
the present meta-analysis and those from these recent treatment for ADHD, in terms of either ADHD symptoms or
EAGG meta-analyses. In terms of effects on ADHD symp- other cognitive correlates. Future research should focus on
toms, neurofeedback, cognitive training, and behavioral the following: identifying the most appropriate electro-
interventions show almost identically sized positive and physiological treatment target; increasing the use of stan-
statically significant effects on total ADHD symptoms scores dard EEG and learning protocols; developing new methods
rated by most proximal assessors (SMD: behavioral to optimize the chances that neurofeedback leads to learning
interventions, 0.35; cognitive training, 0.37; neurofeedback, at the brain level; and identifying predictors of treatment
0.35). Furthermore, in all 3 cases, the effects drop substantially response for individual patients or at least in distinctive
to nonsignificant levels when probably blinded outcomes are subgroups of children. &
used. This is more evident for behavioral interventions where
the effects drop to 0 but is also substantial for neurofeedback Accepted March 28, 2016.
and cognitive training (SMD ¼ 0.15 and 0.20, respectively). Drs. Cortese, Ferrin, Stevenson, and Sonuga-Barke are with the Academic Unit
However, 1 quite striking difference among the 3 meta- of Psychology, Developmental Brain-Behaviour Laboratory, University of
analyses is that both behavioral interventions and cognitive Southampton, Southampton, UK. Dr. Cortese is also with the New York Uni-
training had predictable positive effects on outcomes other versity Child Study Center, New York, and Solent NHS Trust, UK. Dr. Ferrin is
also with the Huntercombe Hospital Maidenhead, Maidenhead, UK. Dr.
than ADHD. For instance, for working memory training, Sonuga-Barke is also with Ghent University, Ghent, Belgium and Aarhus
there were highly significant effects on neuropsychological University, Aarhus, Denmark. Dr. Brandeis and Mr. Aggensteiner are with
measures of working memory, whereas behavioural in- Central Institute of Mental Health, Medical Faculty Mannheim, Heidelberg
University, Mannheim, Germany. Dr. Brandeis is also with Psychiatric Hospi-
terventions improved parenting rated by independent ob- tal, University of Zurich, the Integrative Human Physiology and the Neurosci-
servers and had positive effects on probably blinded ence Center Zurich, University of Zurich, Switzerland, and ETH Zurich. Dr.
measures of conduct problems. By contrast, we did not find Holtmann is with the LWL-University Hospital for Child and Adolescent Psy-
chiatry, Ruhr University Bochum, Germany. Dr. Daley is with the School of
evidence for effects of neurofeedback on neuropsychological
Medicine and MindTech Institute of Mental Health, University of Nottingham,
outcomes. UK. Drs. Santosh, Simonoff, and Stringaris are with the Institute of Psychiatry,
Caution is required when interpreting these findings, Psychology and Neurosciences, King’s College London, and the Maudsley
given a number of limitations in addition to the issues Hospital, London, UK.
raised with regard to the nature of the trials above. First, European ADHD Guidelines Group (EAGG) Members: Phil Asherson, MBBS,
MRPsych, PhD, Tobias Banaschewski, MD, PhD, Daniel Brandeis, PhD, Jan
effect size estimates may be inflated because of the fail- Buitelaar, MD, PhD, David Coghill, MD, Samuele Cortese, MD, PhD, David
ure to report intention to treatment analyses in most Daley, PhD, Marina Danckaerts, MD, PhD, Ralf W. Dittmann, MD, PhD,
trials. Second, there were insufficient trials measuring Manfred D€ opfner, PhD, Maite Ferrin, MD, PhD, Chris Hollis, MD, PhD, Martin
important outcomes such as working memory task per- Holtmann, MD, PhD, Eric Konofal, MD, PhD, Michel Lecendreux, MD, Aribert
Rothenberger, MD, Paramala Santosh, MD, PhD, Joseph A. Sergeant, PhD,
formance, academic skills, general functional impairment, Emily Simonoff, MD, Edmund J. Sonuga-Barke, PhD, Cesar Soutullo, MD,
IQ, and other mental health problems such as conduct HansChristoph Steinhausen, MD, PhD, Jim Stevenson, PhD, Argyris Stringaris,
problems. Third, few trials included long-term outcomes MD, PhD, MRCPsych, Eric Taylor, MD, Saskia van der Oord, PhD, Ian Wong,
PhD, and Alessandro Zuddas, MD.
(Table 1) to allow an evaluation of the extent to which
Support for meetings and analyses was received from Brain Products GMBH,
effects on clinical symptoms grew over time or effects on
Janssen-Cilag, Eli Lilly and Co., Medice, Shire, and Vifor. No honoraria were
neuropsychological processes persisted. Fourth, no trial received, and funders had no input into the review and meta-analysis process
recruited participants based on the presence of ADHD- or the writing of the paper.
related deficits in EEG signature (i.e., altered theta–beta Drs. Cortese and Ferrin are joint first authors.
ratio). This may limit the chance for neurofeedback- The authors thank L. Eugene Arnold, MD, MEd, Ohio State University, USA,
related improvements in symptoms. Fifth, it was neces- and Hanna Christiansen, PhD, University of Marburg, Germany, for providing
sary to pool data from diverse measures from different additional information on studies. The authors also acknowledge the
constructive comments of the referees.
tasks to have sufficient trials to analyze neuropsycho-
Disclosure: Dr. Cortese has received grant or research support from the Solent
logical functions. Although in principle the pooled National Health Service (NHS) Trust, UK. He has received honorarium and
measures tapped the same neuropsychological domain, travel expenses from the Association for Child and Adolescent Mental Health
this inevitably was a somewhat arbitrary process that (ACAMH), UK. Dr. Ferrin was invited to participate to give a workshop on
“Behavioural Interventions in ADHD” at the Meeting of Minds, Stockholm
likely increased SMD heterogeneity. In addition, ADHD 2015, organized by Shire. Dr. Brandeis has served as an unpaid scientific
is a pathophysiologically heterogeneous disorder, and advisor for a European Union (EU)-funded neurofeedback trial. Dr. Holtmann
distinct EEG subtypes have been described (e.g., cortical has served in an advisory or consultancy role for Medice and Shire, and has
hyperarousal versus hypoarousal subtypes). Patients received conference attendance support or was paid for public speaking by Eli
Lilly and Co., Medice, Neuroconn, and Shire. Dr. Daley has provided
might require more specific and tailored training target- educational talks for Eli Lilly and Co. and Shire, has attended an advisory
ing different deficits associated with ADHD. Combining

JOURNAL OF THE AMERICAN ACADEMY OF C HILD & ADOLESCENT PSYCHIATRY

VOLUME 55 NUMBER 6 JUNE 2016 www.jaacap.org 453
 CORTESE et al.

board for Eli Lilly and Co., has received support for educational travel from Eli Dr. Sonuga-Barke has received research funding and conference support from
Lilly and Co., Shire, and HP Pharma, and research funding from Shire. He has Shire Pharma; has received speaker fees from Shire Pharma, Janssen-Cilag,
received royalties from the sale of a self-help version of the New Forest and Medice; has received book royalties from Oxford University Press and
Parenting Programme. Dr. Santosh has received research funding from the EU Jessica Kingsley; has served as a consultant to Shire Pharma, Neurotech
(FP7 Programme) and the National Institute for Health Research (NIHR). He is Solutions, Aarhus University, Copenhagen University, and KU Leuven.
a director and shareholder of HealthTracker, Ltd., UK, and HighStreet Med- Dr. Stevenson and Mr. Aggensteiner report no biomedical financial interests
ical Dental, Gillingham, UK. Dr. Simonoff has received grant funding from the or potential conflicts of interest.
following sources: NIHR; Autism Speaks; European Innovative Medicines
Initiative; Wellcome Trust; South London and Maudsley Charitable Founda- Correspondence to Edmund J.S. Sonuga-Barke, PhD, Developmental Brain-
tion; Biomedical Research Centre Nucleus Pilot Funding; NIHR Evaluation, Behaviour Laboratory, Psychology Academic Unit, University of South-
Trials and Coordinating Centre (NETSCC). Dr. Stringaris has received grant ampton, University Road, Southampton SO17 1BJ, UK; e-mail: ejb3@soton.
or research support from the Guy’s and St. Thomas’ Charity, the Wellcome ac.uk
Trust, NIHR, and University College London for a joint project with Johnson 0890-8567/$36.00/ª2016 American Academy of Child and Adolescent
and Johnson. He has received royalties from Cambridge University Press for Psychiatry
The Maudsley Reader in Phenomenological Psychiatry and Oxford University
Press for Disruptive Mood: Irritability in Children and Adolescents. http://dx.doi.org/10.1016/j.jaac.2016.03.007

REFERENCES
1. Faraone S, Asherson P, Banaschewski T, et al. Attention-deficit/hyper- 17. Sonuga-Barke EJ, Brandeis D, Cortese S, et al. Nonpharmacological in-
activity disorder. Nature Reviews: Disease Primers. 2015;1:15020. terventions for ADHD: systematic review and meta-analyses of ran-
2. Polanczyk GV, Salum GA, Sugaya LS, Caye A, Rohde LA. Annual domized controlled trials of dietary and psychological treatments. Am J
research review: a meta-analysis of the worldwide prevalence of mental Psychiatry. 2013;170:275-289.
disorders in children and adolescents. J Child Psychol Psychiatry. 2015; 18. Micoulaud-Franchi JA, Geoffroy PA, Fond G, Lopez R, Bioulac S,
56:345-365. Philip P. EEG neurofeedback treatments in children with ADHD: an
3. Taylor E. Uses and misuses of treatments for ADHD. The second Birgit updated meta-analysis of randomized controlled trials. Front Hum
Olsson lecture. Nord J Psychiatry. 2014;68:236-242. Neurosci. 2014;8:906.
4. Cortese S, Ferrin M, Brandeis D, et al. Cognitive training for attention- 19. Arns M, Heinrich H, Strehl U. Evaluation of neurofeedback in ADHD:
deficit/hyperactivity disorder: meta-analysis of clinical and neuropsy- the long and winding road. Biol Psychol. 2014;95:108-115.
chological outcomes from randomized controlled trials. J Am Acad Child 20. Higgins JP, Green S, eds. Cochrane Handbook for Systematic Reviews of
Adolesc Psychiatry. 2015;54:164-174. Interventions. Chichester: John Wiley and Sons; 2011.
5. Daley D, Van der Oord S, Ferrin M, et al. Behavioral interventions in 21. RevMan, version 5.1. Copenhagen, Nordic Cochrane Centre, Cochrane
attention-deficit/hyperactivity disorder: a meta-analysis of randomized Collaboration, 2011.
controlled trials across multiple outcome domains. J Am Acad Child 22. Morris SB. Estimating effect sizes from pretest-posttest-control group
Adolesc Psychiatry. 2014;53:835-847. designs. Organ Res Methods. 2008;11:364-386.
6. Fabiano GA, Schatz NK, Aloe AM, Chacko A, Chronis-Tuscano A. 23. StataCorp. 2013. Stata Statistical Software: Release 13. College Station,
A systematic review of meta-analyses of psychosocial treatment for TX: StataCorp LP.
attention-deficit/hyperactivity disorder. Clin Child Fam Psychol Rev. 24. Linden M, Habib T, Radojevic V. A controlled study of the effects of
2015;18:77-97. EEG biofeedback on cognition and behavior of children with attention
7. Stevenson J, Buitelaar J, Cortese S, et al. Research review: The role of diet deficit disorder and learning disabilities. Biofeedback Self Regul. 1996;
in the treatment of attention-deficit/hyperactivity disorder—an appraisal 21:35-49.
of the evidence on efficacy and recommendations on the design of future 25. Holtmann M, Grasmann D, Cionek-Szpak E, et al. Specific effects of
studies. J Child Psychol Psychiatry. 2014;55:416-427. neurofeedback on impulsivity in ADHD. Kindheit Entwicklung. 2009;18:
8. Arns M, Heinrich H, Ros T, Rothenberger A, Strehl U. Editorial: 95-104.
Neurofeedback in ADHD. Front Hum Neurosci. 2015;9:602. 26. Arnold LE, Lofthouse N, Hersch S, et al. EEG neurofeedback for ADHD:
9. Holtmann M, Sonuga-Barke E, Cortese S, Brandeis D. Neurofeedback for double-blind sham-controlled randomized pilot feasibility trial. J Atten
ADHD: a review of current evidence. Child Adolesc Psychiatr Clin North Disord. 2013;17:410-419.
Am. 2014;23:789-806. 27. Bakhshayesh AR, Hansch S, Wyschkon A, Rezai MJ, Esser G. Neuro-
10. Lubar JF, Shouse MN. EEG and behavioral changes in a hyperkinetic feedback in ADHD: a single-blind randomized controlled trial. Eur Child
child concurrent with training of the sensorimotor rhythm (SMR): a Adolesc Psychiatry. 2011;20:481-491.
preliminary report. Biofeedback Self Regul. 1976;1:293-306. 28. Beauregard M, Levesque J. Functional magnetic resonance imaging
11. Arns M, Conners CK, Kraemer HC. A decade of EEG theta/beta investigation of the effects of neurofeedback training on the neural bases
ratio research in ADHD: a meta-analysis. J Atten Disord. 2013;17: of selective attention and response inhibition in children with attention-
374-383. deficit/hyperactivity disorder. Appl Psychophysiol Biofeedback. 2006;
12. Holtmann M. Commentary: Persistent time estimation deficits in ADHD? 31:3-20.
From developmental trajectories to individual targets for intervention— 29. Bink M, van NC, Popma A, Bongers IL, van Boxtel GJ. Behavioral effects
reflections on Doehnert et al. (2013). J Child Psychol Psychiatry. 2013;54: of neurofeedback in adolescents with ADHD: a randomized controlled
271-272. trial. Eur Child Adolesc Psychiatry. 2015;24:1035-1048.
13. Doehnert M, Brandeis D, Schneider G, Drechsler R, Steinhausen HC. 30. Christiansen H, Reh V, Schmidt MH, Rief W. Slow cortical potential
A neurophysiological marker of impaired preparation in an 11-year neurofeedback and self-management training in outpatient care for
follow-up study of attention-deficit/hyperactivity disorder (ADHD). children with ADHD: study protocol and first preliminary results of a
J Child Psychol Psychiatry. 2013;54:260-270. randomized controlled trial. Front Hum Neurosci. 2014;8:943.
14. Nigg JT, Lewis K, Edinger T, Falk M. Meta-analysis of attention-deficit/ 31. Gevensleben H, Holl B, Albrecht B, et al. Is neurofeedback an efficacious
hyperactivity disorder or attention-deficit/hyperactivity disorder symp- treatment for ADHD? A randomised controlled clinical trial. J Child
toms, restriction diet, and synthetic food color additives. J Am Acad Psychol Psychiatry. 2009;50:780-789.
Child Adolesc Psychiatry. 2012;51:86-97. 32. Heinrich H, Gevensleben H, Freisleder FJ, Moll GH, Rothenberger A.
15. Arns M, de RS, Strehl U, Breteler M, Coenen A. Efficacy of neurofeedback Training of slow cortical potentials in attention-deficit/hyperactivity
treatment in ADHD: the effects on inattention, impulsivity and hyper- disorder: evidence for positive behavioral and neurophysiological ef-
activity: a meta-analysis. Clin EEG Neurosci. 2009;40:180-189. fects. Biol Psychiatry. 2004;55:772-775.
16. Fabiano GA, Pelham WE, Jr., Coles EK, Gnagy EM, Chronis- 33. Maurizio S, Liechti MD, Heinrich H, et al. Comparing tomo-
Tuscano A, O’Connor BC. A meta-analysis of behavioral treatments for graphic EEG neurofeedback and EMG biofeedback in children
attention-deficit/hyperactivity disorder. Clin Psychol Rev. 2009;29: with attention-deficit/hyperactivity disorder. Biol Psychol. 2014;
129-140. 95:31-44.

JOURNAL OF THE AMERICAN ACADEMY OF C HILD & ADOLESCENT PSYCHIATRY

454 www.jaacap.org VOLUME 55 NUMBER 6 JUNE 2016
 META-ANALYSIS OF NEUROFEEDBACK OUTCOMES IN ADHD

34. Steiner NJ, Sheldrick RC, Gotthelf D, Perrin EC. Computer-based 41. Zuberer A, Brandeis D, Drechsler R. Are treatment effects of neuro-
attention training in the schools for children with attention deficit/ feedback training in children with ADHD related to the successful
hyperactivity disorder: a preliminary trial. Clin Pediatr (Phila). 2011;50: regulation of brain activity? A review on the learning of regulation of
615-622. brain activity and a contribution to the discussion on specificity. Front
35. Steiner NJ, Frenette EC, Rene KM, Brennan RT, Perrin EC. Neuro- Hum Neurosci. 2015;9:135.
feedback and cognitive attention training for children with attention- 42. Vernon D, Frick A, Gruzelier J. Neurofeedback as a treatment for ADHD:
deficit hyperactivity disorder in schools. J Dev Behav Pediatr. 2014; a methodological review with implications for future research.
35:18-27. J Neurother. 2004;8:53-82.
36. Vollebregt MA, van Dongen-Boomsma M, Buitelaar JK, Slaats- 43. Steiner NJ, Frenette EC, Rene KM, Brennan RT, Perrin EC. In-school
Willemse D. Does EEG-neurofeedback improve neurocognitive func- neurofeedback training for ADHD: sustained improvements from a
tioning in children with attention-deficit/hyperactivity disorder? A randomized control trial. Pediatrics. 2014;13:483-492.
systematic review and a double-blind placebo-controlled study. J Child 44. Bink M, van Nieuwenhuizen C, Popma A, Bongers IL, van Boxtel GJ.
Psychol Psychiatry. 2014;55:460-472. Neurocognitive effects of neurofeedback in adolescents with ADHD: a
37. Gioia GA. BRIEF: Behavior Rating Inventory of Executive Function: randomized controlled trial. J Clin Psychiatry. 2014;75:535-542.
Professional Manual. Odessa, FL: Psychological Assessment 45. van Dongen-Boomsma M, Vollebregt MA, Slaats-Willemse D,
Resources; 2000. Buitelaar JK. A randomized placebo-controlled trial of electroencepha-
38. Liechti MD, Valko L, Muller UC, et al. Diagnostic value of resting elec- lographic (EEG) neurofeedback in children with attention-deficit/
troencephalogram in attention-deficit/hyperactivity disorder across the hyperactivity disorder. J Clin Psychiatry. 2013;74:821-827.
lifespan. Brain Topogr. 2013;26:135-151. 46. Wangler S, Gevensleben H, Albrecht B, et al. Neurofeedback in children
39. Monastra VJ, Lubar JF, Linden M, et al. Assessing attention deficit with ADHD: specific event-related potential findings of a randomized
hyperactivity disorder via quantitative electroencephalography: an initial controlled trial. Clin Neurophysiol. 2011;122:942-950.
validation study. Neuropsychology. 1999;13:424-433. 47. Arns M, Strehel U. Evidence for efficacy of neurofeedback in ADHD? Am
40. Heinrich H, Busch K, Studer P, Erbe K, Moll GH, Kratz O. EEG spectral J Psychiatry. 2013;170:799-800.
analysis of attention in ADHD: implications for neurofeedback training? 48. Sonuga-Barke E, Brandeis D, Cortese S, et al. Response to Chronis-
Front Hum Neurosci. 2014;8:611. Tuscano, et al. and Arns and Strehel. Am J Psychiatry. 2013;170:800-802.

JOURNAL OF THE AMERICAN ACADEMY OF C HILD & ADOLESCENT PSYCHIATRY

VOLUME 55 NUMBER 6 JUNE 2016 www.jaacap.org 455