Content uploaded by Ute Strehl
Author content
All content in this area was uploaded by Ute Strehl on Jun 16, 2014
Content may be subject to copyright.
209
International Journal of Bioelectromagnetism www.ijbem.org
Vol. 10, No. 4, pp.209-232, 2008
Long term effects after feedback of slow cortical
potentials and of theta-beta-amplitudes in children with
attention-deficit/hyperactivity disorder (ADHD)
Cihan Gania, , Niels Birbaumera, Ute Strehla
a Institute of Medical Psychology and Behavioral Neurobiology
Correspondence: Ute Strehl, Institute of Medical Psychology and Behaioral Neurobiology, University of Tübingen,
Gartenstr. 29, 72074 Tübingen, Germany. E-mail: ute.strehl@uni-tuebingen.de
Abstract. Though it had already been shown in the 1970s that neurofeedback improves attention, academic
performance and social behavior in children with ADHD, it has not been considered as a standard therapy
so far. This is mainly due to the small number of controlled studies fulfilling methodological standards -
especially long term data was not available so far. We are the first to present long term data of children
undergoing neurofeedback training. 47 patients in the age of 8 – 12 years were randomly assigned to two
different training groups. One group was trained to self regulate slow cortical potentials (SCP), the other
group tried to influence Theta- and Beta-amplitudes. Follow-up evaluation was carried out 6 months and
more than 2 years after the last training session. Eleven children of the SCP group and 12 children of the
Theta/Beta group took part in three booster sessions. Parents rated behavioral symptoms as well as
frequency and impact of problems. Attention was measured with the Testbatterie zur
Aufmerksamkeitsprüfung (TAP).All improvements in behavior and attention that had been observed at
previous assessments turned out to be stable. Yet another significant reduction of number of problems and
significant improvement in attention was observed. EEG-self regulation skills were preserved. In each
group, half of the children no longer met ADHD - criteria. Neurofeedback appears to be an alternative or
complement to traditional treatments. The stability of changes might be explained by normalizing of brain
functions that are responsible for inhibitory control, impulsivity and hyperactivity.
Keywords: Neurofeedback; ADHD; Slow cortical potentials (SCPs); Theta/Beta – ratio; long term effects.
1. Introduction
Though stimulant medication is regarded as the most effective therapy for attention deficit
hyperactivity disorder (ADHD), till recently there were no studies demonstrating evidence for long-term
benefits from pharmacotherapy [Goldman, Genel et al., 1998]. Children, who discontinued medication,
experience a considerable loss of improvement at follow-up. Only lately the “3- year follow-up of the
NIMH MTA Study” [Jensen, Arnold et al., 2007] showed that early advantages of medication management
compared with behavioral therapy and community care were no longer present at 3- year follow-up. In
addition, a large number of non-responders - approximately 25% [DuPaul and Connor, 1998] - and side
effects of stimulant medication, such as several vegetative complaints [Schachter, Pham et al., 2001] and
reduced growth [Swanson, Elliott et al., 2007] have led to an increased demand for alternative therapy
options .
210
Neurofeedback has been used as a treatment for attention deficit hyperactivity disorder since the
1970s [Lubar and Shouse, 1976] when scientist recognized, that children with seizure disorders improved
at school when they were treated with neurofeedback. In the beginning neurofeedback was dismissed as
poorly researched and overly hyped, yet several studies have shown that after neurofeedback behavioral
problems, attention, academic performance, and social behavior are improved [Lubar and Lubar, 1984].
Neurofeedback protocols mainly aim at two different electrophysiological patterns. Children with
ADHD show increased Theta and decreased Beta activity in their spontaneous EEG [Monastra, Lubar et al.,
1999] and event- related potentials in children with ADHD are characterized by decreased amplitudes and
prolonged latencies compared to healthy children [Johnstone, Barry et al., 2001].
Although several studies had shown that feedback of EEG frequencies leads to improvement of
symptoms neurofeedback has not been acknowledged because of methodological shortcomings (e.g. no
randomization, no EEG data, no controls, no follow-up [Ramirez, Desantis et al., 2001] ). Since 2002 this
has changed to the better. Two controlled studies showed that neurofeedback of spontaneous EEG activity
(e.g. theta and beta) leads to the same improvements in behavior and academic performance as medication
[Fuchs, Birbaumer et al., 2003] and that improvement after a therapy combining medication, parental
counseling and individual school counseling endured after washout of medication only if neurofeedback
had been added [Monastra, Monastra et al., 2002].
Self regulation training of slow cortical potentials aims at the deviant latencies and amplitudes of
event related potentials. Slow cortical potentials (SCPs) are a special type of event related potentials
reflecting the excitation threshold of the upper cortical layer. They are slow direct current shifts. SCP shifts
in the electrical negative direction reflect a reduction of the excitation threshold while shifts in the electrical
positive direction reflect an increase of the excitation threshold [Rockstroh, Elbert et al., 1989]. Children
with attentional problems have an impaired ability to regulate their SCPs [Rockstroh, Elbert et al., 1990].
There is only a small number of studies aiming at self-regulation of SCPs [Heinrich, Gevensleben et al.,
2004] [Strehl, Leins et al., 2006]. [Heinrich, Gevensleben et al., 2004] showed a reduction of ADHD
symptoms by 25 % after training of slow cortical potentials with neurofeedback. [Drechsler, Straub et al.,
2007] showed that children with a diagnosis of ADHD who underwent neurofeedback training of slow
cortical potentials improved more than children who had participated in cognitive behavior group therapy.
Yet the vast majority of studies so far has examined self-regulation training of
electroencephalogram (EEG) frequency bands. In most cases the training rationale with ADHD patients
was to decrease activity in the Theta band and to increase activity in the Beta band.
In a randomized study neurofeedback of Theta/Beta Frequencies or SCPs, [Leins, Goth et al., 2007]
treated children between 8 and 13 years with either with a Theta / Beta or a SCP protocol. They
demonstrated that children gained the ability to self regulate cortical activity. Moreover patients improved
significantly in attention and IQ, as well as in behavior. The results remained constant six months after the
end of training. For the current study the same population was re-invited for the 2 year follow-up. We
present the first randomized long-term follow-up study, providing EEG data from neurofeedback sessions 2
211
years after the end of treatment. Furthermore behavioral ratings have been assessed and associated with the
EEG data.
The goal of the 2 year follow-up is to determine a) whether patients kept the ability to self regulate
cortical activation, b) whether improvements in attention and behavior remained stable and c) whether the
different treatments lead to differences in the stability of cortical self regulation and clinical effects.
2. Methods
2.1. Study Design and Participants
Figure 1 shows the schedule of the whole study and the number of children participating at each
assessment point. 23 children participated in the 2 years follow-up assessment, 11 out of the SCP-Group
and 12 of the Theta/Beta group. For our study the assessment points “screening”, “end of treatment” and “2
year follow-up” were included. Detailed comparisons between screening, end of treatment and six month
follow-up have been reported by [Strehl, Leins et al., 2006] and [Leins, Goth et al., 2007]. Due to the
earlier time of data analysis in [Strehl, Leins et al., 2006] and [Leins, Goth et al., 2007], sample sizes in
those articles might deviate from the ones given in Figure 1.
212
Figure 1. Study design and schedule (SCP = Slow cortical potentials. T/B = Theta/Beta ratio.) Note: Only patients
who completed the first 30 sessions are included in the graph. Drop-out rate between “Screening” and “End of
training”: 2 in the SCP and 3 in the Theta/Beta group.
Participants were recruited from the outpatient clinic for psychotherapy at the University of
Tübingen and from local psychiatric practitioners and psychotherapists. Approval was obtained from the
local ethics committee of the faculty of medicine according to the convention of Helsinki.
Each child met DSM-IV criteria for ADHD inattentive, hyperactivity or combined type and was aged
between 8 and 13. Children with a Full-Scale IQ of <80 or additional neurological conditions were
excluded. After parents and children had signed written informed consent, children were randomly assigned
to treatment. Children were matched according to age, sex, IQ, diagnosis and medication. Parents and
children were blinded according to group assignment (Theta/Beta or SCP Feedback).
In order to evaluate long-term outcome, self regulation skills, behavioral symptoms and variables
of attention were assessed as follows:
- 3 sessions EEG Feedback
- DSM IV questionnaire for parents
- Eyberg Child Behavior Inventory [Eyberg, 1999]
- Conners’ Rating Scale (Translated into German) [Conners, 1997]
213
- Testbatterie zur Aufmerksamkeitsprüfung, Version 1.7 [Zimmermann, 2002]
At preceding assessment points following instruments had been used in addition:
- Kindl-Questionnaire for Measuring Health-Related Quality of Life in Children and Adolescents,
parents’ and children’s version [Ravens-Sieberer, 2003]
- German Version of Wechsler Intelligence Scale for Children: Hamburg-Wechsler-Intelligenztest
für Kinder [Tewes, 1999]
- DSM-IV – questionnaire for teachers to assess DSM-IV-criteria for ADHD.
Because of possible practice effects the Intelligence Scale was not applied at the 2 year follow-up.
As most children’s teachers had changed between screening and 2 year follow-up, and because parents did
not want to involve school any more (especially if symptoms had vanished), it was decided to do without
teachers’ ratings.
Parents whose children did not participate at the 2 year follow-up were interviewed by telephone
and were asked for their reasons to decline the invitation.
2.2. Neurofeedback sessions in the 2-year follow-up
Setting, instruments and trainer were identical in both groups. The only but fundamental
difference between both groups consisted in the feedback signal. Although detailed descriptions have been
given elsewhere [Strehl, Leins et al., 2006; Leins, Goth et al., 2007] basic information about the two
different training modalities will be repeated here.
Training of slow cortical potentials
Slow cortical potentials can be observed with a latency of about 500ms after stimulus onset and
may endure several seconds. They reflect the excitability of brain areas. Electrically negative SCPs
decrease the excitation threshold while positive SCPs increase this threshold. It is assumed that developing
self regulation skills will help to improve the symptoms of hypoarousal in children with ADHD.
Children were seated in an armchair in front of a 17’’ monitor in a distance of approximately 50
inches. Electrodes were affixed on Cz, and on both mastoids, with a 10 kΩ resistance between Cz and the
mastoids. EEG signals were amplified using an EEG 8 (Contact Precision Instruments, Cambridge, MA)
amplifier with a time constant of 16s and a low frequency filter of 40 Hz. EEG signals were digitized with a
sampling rate of 256 Hz. The slow-wave amplitude just before the beginning of the active phase of a trial
was taken as baseline and set to zero. For online and offline correction of eye movements a vertical
electrooculogram was recorded with electrodes fixed above and below the left eye.
214
During the active phase the slow-wave amplitude was calculated every 62.5 milliseconds as the
average of the preceding 500 milliseconds. The difference between every 500 millisecond amplitude in the
active phase and the amplitude during the baseline corresponded to the position of the feedback signal, a
yellow “ball”.
The patients saw two rectangles, one on the top and one on the bottom of the screen. A highlighted
upper rectangle indicated that a SCP shift in the electrical negative direction was demanded; a highlighted
lower rectangle indicated that a positive SCP shift was required (see Figure 2 for a picture of the screen).
Figure 2. Screens: The upper screens depict the beginning (left) and at the end (right) of a trial. The lower screens
indicate a transfer trial at the beginning (left) and at the end (right) of a trial (Figure taken from Strehl et al, 2006
[Strehl, Leins et al., 2006])
Each session consisted of 3 to 5 runs, each run included 39 trials. The ratio between negativity and
positivity tasks was 3:1. Each trial consisted of a 2 second passive phase, followed by a 6 second active
phase. Visual feedback was given by a round yellow circle (“ball” with a 1 inch diameter) that moved
upwards (negativity – so called activation) or downwards (positivity – so called deactivation)
proportionally to the cortical shift. The ball moved from left to right with a constant velocity and up and
down dependent of the amplitude of the cortical shifts. After successful trials a smiley face appeared on the
screen. Furthermore, auditory feedback was given by piano tones with high frequency (negativation) and
low frequency (positivation). A harmonious jingle at the end of a trial signalized success. Depending from
the number of correct results children received a voucher of a certain amount. This voucher could be
exchanged in a local toyshop. The value was a minimum of 3 € and a maximum of 10 € for all sessions.
215
In order to facilitate practice of self regulation in daily life transfer trials were inserted (see lower
panel of Figure 2). No auditory or visual feedback was given during the active phase, only the appearance
of the smiley face and the jingle at the end of a trial delivered information about success.
During the sessions the trainer sat next door and observed the EEG online on a monitor and the
child on a second monitor. Intervention was possible by a two-way intercom or by joining the child.
Training of Theta/Beta ratio
Number of runs and sessions, relation of tasks (activation / deactivation), conditions (feedback /
transfer) as well as the depiction of the feedback signal on the screen and reinforcement was identical to the
SCP-training. Different from the SCP protocol, electrodes were fixed at Cz, and additionally at C3f, which
is halfway between C3 and F3 and C4f, which is halfway between C4 and F4. Electrodes on both mastoids
were used as reference.
The feedback signal was calculated online as the averaged Theta/Beta ratio measured the averaged
activity at C3f and C4f minus the averaged ratio measured at the mastoids. Theta range was between 3 and
7 Hz and Beta range was between 12 and 20 Hz. Baseline phase was expanded to 2 seconds and feedback
phase to 7.5 seconds, as Theta and Beta frequencies show greater oscillations than SCP. An 8 second “pre-
baseline” was measured at the beginning of each training session. Integration of the Theta/Beta ratio
measured during the pre-baseline and the Theta/Beta ratio measured prior to the first trial delivered an
“overall-baseline”, which acted as a reference for the first trial. This reference was corrected online by each
new trial-baseline-ratio during the training session.
So called “activation tasks” demanded a decrease of the Theta/Beta ratio, which could be reached by
either decreasing Theta and / or increasing Beta. So called “deactivation tasks” demanded an increase of the
Theta/Beta ratio. The movement of the ball in the vertical direction reflected the current Theta/Beta ratio.
An increasing ratio resulted in a ball movement downwards, a decreasing ratio in a movement upwards. A
trial was rated as correct when in comparison to the overall-baseline-ratio the averaged ratio was lower in
“activation tasks” and higher in “deactivation tasks”.
SCP
group T/B group
Feedback signal < 1 Hz Theta (3-7 Hz) / Beta (7-20Hz)
Electrodes Cz,
mastoids Cz, C3f, C4f, mastoids
Baseline phase 2 seconds Continuous integration of 8 seconds pre-baseline (taken
before each session) and 2 seconds (taken before each trial)
Feedback phase 6 seconds 7.5 seconds
Window of Feedback
phase considered for
analysis
3-6
seconds
3-7.5 seconds
Table 1: Differences between SCP and T/B protocols
216
2.3. Data analysis
As can be seen from Figure 1 the design of the whole study contained 4 assessment points. Since 6
month-follow-up data are published elsewhere [Leins et al., 2007], here we compare data from screening,
end of training and 2 year follow-up.
EEG Data
According to our goal to determine long-term effects of SCP- and Theta/Beta-Feedback as a therapy
for children with ADHD we investigated whether
a) subjects of both groups were still able to differentiate between activation and deactivation tasks at
2-year follow-up
b) the difference between activation and deactivation tasks changed with time (“screening” =
sessions 2 + 3; “end of treatment” = sessions 29 + 30, “2 year follow-up” = sessions 35 + 36).
Sessions 1 and 34 were discarded as habituation sessions.
SCP - Group
For each child mean SCP amplitudes were calculated for both tasks (positivity / negativity) and
conditions (feedback / transfer) after testing for normal distribution with the Kolmogorov-Smirnov-Test for
each of the three assessment points “screening”, “end of treatment” and “2 year follow-up”).
The differences of SCP amplitudes between both tasks were analyzed for both conditions and each
assessment point by an independent samples t-test. An analysis of variance (ANOVA) with repeated
measures (time x condition x task) was performed in order to evaluate changes between assessment points.
Bonferroni-corrected post hoc paired samples tests were performed in case of a significant result in the
ANOVA. The ANOVA was corrected with Greenhouse-Geisser.
Theta / Beta - Group
The difference between the Theta/Beta ratio during the baseline phase and the ratio during the active
phase was calculated for each subject, for both conditions (feedback/transfer), both tasks
(activation/deactivation) and each assessment point (“screening”, “end of treatment” and “2 year follow-
up”).
After testing for normal distribution with the Kolmogorov-Smirnov-Test the difference between the
Theta/Beta ratio during activation and deactivation tasks was analyzed separately for feedback und transfer
tasks for each assessment point with a one samples t- test.
In order to assess possible changes of the Theta/Beta ratios an ANOVA with repeated measures
(“screening”, “end of treatment” and “2 year follow-up”) was performed for both conditions and task
217
(activation/deactivation). Bonferroni corrected post hoc paired samples test were applied if significant
changes with time were found.
Psychometric test data
After testing for normal distribution, test data and questionnaires were analyzed by a mixed design
ANOVA with repeated measures factors and the group factor SCP vs Theta / Beta in order to examine the
effects of time (“screening”, “end of treatment” and “2 year follow-up”). In case of a significant result in
the ANOVA post hoc paired samples tests were performed.
Effect sizes
Cohen’s d [Cohen, 1988] was used in order to assess effect sizes for each significant result after
correction with Bonferroni. Cohen’s d is equal to the differences between the means M1 - M2, divided by
the pooled standard deviation pooled = [( 1²+ 2²) / 2].
3. Results
3.1. Participants
Groups had been matched for age and IQ. See Table 2 for demographic data and diagnosis. A
comparison between the participants at 2-year follow-up still yielded no significant difference regarding to
IQ and age.
218
SCP Med (mg) IQ Diagosis Screening Diagosis 2 year follow up
Code Age Sex Screening Follow up Screening Hyperactivity Inattention Hyperactivity Inattention
12 8 m 0 0 85 y y n n
13 9 m 40 40 123 y y y y
19 8 f 0 0 98 y y y y
23 8 m 0 0 89 y y y n
24 8 f 0 0 126 y y y y
26 9 m 0 0 116 y y n n
30 9 m 25 20 120 y y n n
31 8 m 0 0 101 y y n n
32 9 m 15 0 111 n y n y
35 10 m 0 0 113 y y n n
40 9 m 0 0 98 n y n y
Theta/Beta 2 10 m 10 0 79 n y n n
6 9 m 15 10 112 n y n n
7 8 m 0 0 99 y n y n
8 8 m 20 5 95 y y n n
25 8 m 0 0 98 y y n n
47 8 m 0 0 100 y y y y
49 8 m 0 0 109 y y y y
58 8 m 0 0 101 y y n y
60 8 f 0 0 105 n y n n
61 12 m 0 0 92 n y n n
67 10 f 20 20 85 y y y y
69 8 m 0 0 93 y y n y
Table 2: Description of the sample before the beginning of treatment and at 2 year follow-up (y = diagnosis, n= no diagnosis).
219
In order to avoid biasing of our data we contacted those parents and children who did not
participate at the follow-up and asked for their reasons. In most cases lack of time was given as a
reason. Full time education was introduced in many schools in Germany at this time.
As can be seen in Table 3 there was no significant difference regarding age and IQ between
children who participated in the 2 year follow-up and those who did not. Yet in the Theta/Beta group a
significant difference in diagnosis at the 6 month follow-up could be observed. Pearson’s 2 test
showed that children who did not participate in the 2 year follow-up had a significantly higher rate of
ADHD diagnosis than children who participated.
SCP group Theta/Beta group
Participants Non-
Participants p Participants Non-
Participants p
IQ (mean) 107,27 98,64 0,076 97,33 100,1 0,49
Age (mean) 8,64 9,86 0,056 8,75 9,7 0,11
Diagnosis
session 2+3
Combined 9 11 0,84 7 9 0,096
Inattentive 2 3 0,84 4 1 0,19
Hyperactive 0 0 1 0 0,35
6 month
folllow-up
Combined 3 7 0,24 2 7 0,011
Inattentive 4 4 0,67 3 1 0,36
Hyperactive 2 2 0,79 1 0 0,35
No diagnosis 2 1 0,39 6 2 0,14
Table 3: Comparison of participating and non-participating children at 2 year follow-up.
3.2. Regulation of SCPs
SCP amplitudes were distributed normally (Kolmogorow-Smirnov). Differences between
activation and deactivation tasks in the feedback condition were close to significance only at the end of
treatment (t 10 =-2,220, p=0.051, ES = 0.39). The difference was not significant at other assessment
points. No significant difference could be observed in the transfer condition.
In the feedback condition the difference between activation and deactivation tasks increased
significantly with time (F 2, 20 = 13.226, p=0.03). In the transfer condition the increase was close to
significance (F 2, 20 = 3.02, p= 0.079). Yet an F-value which is higher than “1” indicates that the p-value
is probably significant in a larger sample. A significant interaction between time x task (positivity /
negativity) could be observed for the feedback condition (F 2, 40 = 4.38, p = 0.032) but not for the
220
transfer condition. In the feedback condition post hoc paired samples tests revealed a significant
increase of the difference between SCP amplitudes in activation and deactivation tasks between
screening and end of treatment (t 10 = 4,22, p = 0.006, ES = 0.53). Between screening and 2-year
follow-up the increase of this difference was close to significance (t 10 = 2,789 p=0,057, ES = 0.40). At
2-year follow-up a significant decrease of the difference between SCP amplitudes in activation and
deactivation tasks in the feedback condition compared to end of treatment could be observed (t 10=-
4.494, p = 0.003, ES = 0.21).
As can be seen in Figure 3 (upper panel) SCP amplitudes in activation tasks (feedback)
increased significantly between screening and end of treatment (t 10 =3,039, p= 0.036, ES =0.47).
Between end of treatment and 2-year follow-up the mean amplitudes in negativity tasks did not change
significantly.
As shown in Figure 3 (both panels), in the transfer condition children were able to produce
positive shifts in positivity tasks for the first time at 2 year follow-up, while in the feedback condition
they already had this skill at the end of training. Yet at following assessment points this ability was no
longer present.
221
2,81
-2,75
-6,44
0,92
-5,80
-2,98
-8
-6
-4
-2
0
2
4
1
Mean SCP amplitude, µV
Negativity tasks, feedback
Positivity tasks, feedback
Sessions 2+3 End of treatment 2
y
ear follow-u
p
p = 0.006
p = 0.057
p = 0.003
p = 0.036
p = 0.051
1,49
-3,50
-6,32
-2,07
-3,49
0,85
-8
-6
-4
-2
0
2
4
1
Mean SCP amplitude, µV
Negativity tasks, transfer
Positivity tasks, transfer
Session 2 + 3 End oftreatment 2
y
ear follow-up
Figure 3. Upper panel: Mean amplitudes in negativity and positivity trials in the feedback condition during
screening, end of treatment and 2 year follow- up. Lower panel: Mean amplitudes in negativity and positivity trials
in the transfer condition during screening, end of treatment and 2 year follow- up.
3.2. Self regulation of Theta/Beta ratios
Theta/Beta ratios were distributed normally (Kolmogorow-Smirnov). Differences between
activation and deactivation tasks in the feedback condition (Figure 4, upper panel) were close to
significance at the end of treatment (t 11 =-2,127, p = 0.057, ES = 0.45). The difference between
activation and deactivation tasks did not change significantly with time neither in the transfer (F 2, 22 =
0.22, p = 0.71) nor in the feedback (F 2, 22 = 0.20, p = 0.69) condition.
222
The increase of the Theta/Beta ratio (see Figure 5, lower panel) in deactivation tasks between
screening and 2-year follow-up was close to significance in the transfer condition (t 11 =-2.536, p =
0.084, ES = 0.31).
No further significant changes between any of the assessment points could be observed.
-0,4
-0,3
-0,2
-0,1
0
0,1
1
Mean Theta/Beta Ratio
Activation tasks, feedback
Deactivation tasks, feedback
Screenin
g
End of trainin
g
2
y
ear follow-up
-0,4
-0,3
-0,2
-0,1
0
0,1
1
Mean, Theta/Beta ratio
Activation tasks, transfer
Deactivation tasks, transfer
Screenin
g
End of 2
y
ear follow-
p = 0.084
Figure 4. Upper panel: Mean Theta / Beta ratios in activation and deactivation tasks in the feedback condition
during screening, end of treatment and 2 year follow- up. Lower panel: Mean Theta / Beta ratios in activation and
deactivation tasks in the transfer condition during screening, end of treatment and 2 year follow- up.
223
3.3. Behavior
Parental ratings
The number of DSM IV criteria for both inattention (F 2, 40 = 16.40, p=0.00) and hyperactivity (F
2, 40 = 14.59, p=0.00) decreased significantly over time. Interaction of group x time and differences
between groups did not reach significance.
As shown in Figure 5, post hoc Wilcoxon Signed ranks tests revealed significant changes for
both groups. In the SCP group a significant decrease of the number of DSM IV criteria for inattention
could be observed between screening and 2 year follow-up (Z = -2.530, p = 0.033, ES = 0.51). The
decrease of the number of DSM IV criteria from screening to 2 year follow-up for hyperactivity in the
SCP group after Bonferroni correction was close to significance (Z = -2.203, p = 0.084, ES = 0.55). In
the Theta/Beta group a significant decrease of DSM IV criteria could be observed for both
hyperactivity (Z = -2.825, p = 0.015, ES = 0.53) and inattention (Z=-2.593, asymp. sig = 0.033, Z =
0.39) between screening and 2 year follow-up.
7,64
6,91
5,60
6,82
6,27
4,50
7,00
5,25
4,83
6,17
5,00
3,92
3
4
5
6
7
8
1
Mean, DSM criteria
Sessions 2 + 3
End of treatment
2 year follow-up
SCP,
Inattention SCP,
Hyperactivity Theta / Beta
Inattention Theta / Beta
Hyperactivity
p = 0.033
p = 0.015
p = 0.033
Figure 5: DSM IV criteria, parents' ratings. Scores below 6 are considered as normal.
Group-means for hyperactivity and inattention in both groups were below the cut-off-value of 6.
In the SCP group only 3 out of 11 children still had a diagnosis of combined ADHD at 2 year follow-
up, 5 no longer had a diagnosis of ADHD at all. In the Theta/Beta group only 3 children still had a
diagnosis of combined ADHD, 6 did not meet DSM criteria anymore. See Table 4 for further details.
224
Screening End of Training 2 year follow-up
Group SCP Theta/Beta SCP Theta/Beta SCP Theta/Beta
Patients 25 22 25 22 11 12
Male 20 18 20 18 8 10
Female 5 4 5 4 3 2
Diagnosis, n
ADHD 20 16 17 7 3 3
ADD, predominantly
inattentive type 5 5 5 5 2 2
ADHD, predominantly
hyperactive type 0 1 2 4 1 1
No Diagnosis of
ADHD
0 0 1 6 5 6
Medication
(Ritalin, 18-60mg) 7 5 7 5 2 3
Table 4: Development of diagnosis assessed by DSM-IV questionnaire.
In both groups Fisher's exact test showed that the changes of diagnosis from screening to 2-year
follow-up are significant (SCP- group: p=0.006, Theta/Beta group: 0.03)
Frequency of problems at home and impact of problems at home were assessed by the Eyberg
questionnaire. A general linear model showed a significant change for both frequency of problems (F 2,
40 = 22.27, p = 0.00) and impact of problems (F 2, 40 = 12.436, p = 0.00). No significant time x group
interaction could be observed. Post hoc paired samples tests revealed a significant decrease of
frequency of problems in the SCP group between screening and 2 year follow-up (t 9 = 5.349, p = 0.00
ES = 0.59) and also for impact of problems (t 9 = 4.263, p = 0.006, ES = 0.34). In the Theta/Beta group
post hoc samples test also showed a significant improvement for frequency of problems between
screening and 2 year follow-up (t 11 = 4.195, p = 0.003, ES = 0.46) and also between end of treatment
and 2 year follow-up (t 11 = 3.252, p = 0.024, ES = 0.23) The impact of problems decreased
significantly in the Theta/Beta group between screening and 2 year follow-up (t 11 = 3.593, p = 0.012,
ES = 0.45). At 2 year follow-up means for frequency of problem in both groups were below the cut-off
value of 126 for the first time (see Figure 6).
225
150,09
132,64
115,90
153,08
138,92
121,42
100
110
120
130
140
150
160
1
Mean, Frequency
Sessions 2 + 3
End of treatment
2 year follow-up
SCP T/B
p = 0.00
p = 0.003
p = 0.024
Figure 6. Frequency of problems, parents' ratings (Eyberg questionnaire); Scores below 127 are considered as
normal.
18,09
17,00
12,40
16,00
13,33
10,00
0
5
10
15
20
1
Impact
Sessions 2 + 3
End of treatment
2 year follow-up
p = 0.006
SCP Theta / Beta
p = 0.012
Figure 7. Impact of problems, parents' ratings (Eyberg questionnaire); Scores below 45 are considered as normal.
By using Connors rating scale parents rated their children's behavior three days in succession at
each assessment point. The changes over time were significant (F 2, 40 = 8.277, p=0.01). No time x
group interaction could be observed. There was no significant difference between the groups. Post hoc
226
tests showed a significant improvement for the SCP group between screening and 2 year follow-up (t 9
= 5.142, p = 0.003, ES = 0.65). As shown in Figure 8, mean values in both groups were below the cut-
off value of 15 at 2 year follow-up.
54,73
44,64
32,20
49,58 48,64
37,25
20
25
30
35
40
45
50
55
60
1
Frequency of problems
Sessions 2 + 3
End of treatment
2 year follow-up
SCP Theta / Beta
p = 0.003
.
Figure 8. Frequency of problems, parents' ratings (Connors questionnaire); Scores below 15 are considered as
normal.
3.4. Attention
The “Testbatterie zur Aufmerksamkeitsprüfung” [Zimmermann, 2002] was used in order to
assess attention. Seven subtests evaluated percent ranges for speed, omissions and commissions. Single
test results were aggregated to the number of results with a below-average achievement (below 25th
percentile) and to the number of results with an above-average achievement (above 75th percentile).
There was a significant decrease by time for below achievements (F 2, 40 = 28.399, p = 0.00) and a
significant increase for above achievements (F 2, 40 = 26.349, p = 0.00). No significant difference
between both groups could be observed neither for above nor for below achievements.
In the SCP-group post-hoc paired samples tests showed a significant decrease of below-average
achievements between screening and 2 year follow-up (t 10= 7.717 p= 0.00 ES= 0.54) and between end
of treatment and 2 year follow-up (t 10 = 4.404 p= 0.003 ES= 0.225). A significant increase of above-
average achievements between screening and 2 year follow-up (t 10 = -6.328 p= 0.00 ES= 0.58) could
be shown, too.
In the Theta / Beta group children showed significantly more above-average achievements (t 10 =
-4.755 p= 0.003 ES= 0.39) at 2 year follow-up compared to screening. Reduction of below-average
227
achievements between screening and 2 year follow-up was close to significance (t 10 = 2.593 p= 0.081
ES = 0.22). In addition a significant increase of above-average achievements between end of treatment
and 2 year follow-up (t 10 = -3.758 p= 0.012 ES= 0.25) could be observed (see Figure 9).
18,27
11,00
6,91
16,67
13,17
11,18 11,18
17,36
22,09
14,58
19,45
12,00
0
5
10
15
20
25
1
Number of sublets (Mean)
Sessions 2 + 3
End of treatment
2 year follow-up
SCP
,
Percentile < 25% T/B
,
Percentile < T/B
,
Percentile > 75%SCP
,
Percentile >
p = 0.00
p = 0.00
p = 0.00
p = 0.003
p = 0.003
p = 0.081
p = 0.012
Figure 9: Comparison of performance in the TAP (Test of attention).
228
SCP Group
p / ES
Screening End of Training 2 year follow up [scr - end] [end - 2yfu] [scr - 2yfu]
DSM Inattention 7.64 ± 1.12 6.91 ± 1.81 5.60 ± 2.11 ns ns 0.033 / 0.51
Hyperactivity 6.82 ± 1.94 6.27 ± 2.00 4.50 ± 1.50 ns ns 0.084 / 0.55
Eyberg Impact 18.09 ± 7.06 17.00 ± 7.82 12.40 ± 8.43 ns ns 0.006 / 0.34
Frequency 150.09 ± 24.87 132.64 ± 32.22 115.90 ± 20.97 0.057 / 0.29 ns 0.000 / 0.59
Connors 54.72 ± 13.60 44.63 ± 23.01 32.20 ± 12.91 ns ns 0.003 / 0.65
TAP Above achievements 11.18 ± 6.97 17.36 ± 10.22 22.09 ± 8.11 0,057 / 0.33 ns 0.000 / 0.58
Below achievements 18.27 ± 9.86 11.00 ± 9.97 6.91 ± 7.57 0.000 / 0.34 0,003 / 0.225 0.000 / 0.54
Theta/Beta Group
Screening End of Training 2 year follow up [scr - end] [end - 2yfu] [scr - 2yfu]
DSM Inattention 7.00 ± 1.28 5.25 ± 1.91 4.83 ± 2.04 0.018 / 0.47 ns 0.015 / 0.53
Hyperactivity 6.17 ± 2.37 5.00 ± 2.56 3.92 ± 2.88 0,051 / 0.256 ns 0,033 / 0.39
Eyberg Impact 16.00 ± 5.11 13.33 ± 6.62 10.00 ± 6.58 ns ns 0,012 / 0.45
Frequency 153.08 ± 26.71 138.92 ± 39.09 121.42 ± 32.85 ns 0,024 / 0.23 0,003 / 0.46
Connors 49.54 ± 18.12 48.63 ± 20.19 37.25 ± 19.81 ns ns ns
TAP Above achievements 12.00 ± 8.75 14.58 ± 10.02 19.45 ± 8.51 ns 0,012 / 0.25 0,003 / 0.39
Below achievements 16.67 ± 12.81 13.17 ± 11.93 11.18 ± 10.70 ns ns 0,081 / 0.22
Table 5. Synopsis of means and p-values in behaviour and attention. (scr = screening prior to the first sessions, end = end of training, 2yfu = 2 year follow-up)
229
4. Discussion
For the first time long-term data after a neurofeedback training of Theta/Beta frequency bands
and SCPs is reported. While cortical self regulation as well as improvements in behavior (parents’ and
teachers’ rating), attention and IQ have been shown to be stable six months after the end of training
[Strehl et al., 2006; Leins et al., 2007], we here present data two years after the end of treatment The
stability of self regulation of EEG as well as of concomitant changes in target behavior and attention
have been analyzed. In both groups self regulation skills and clinical outcome are stable at 2 year
follow-up.. In some cases even an additional improvement at 2 year follow-up could be observed.
EEG self regulation skills showed a mixed pattern regarding groups and conditions. In the SCP
group children gained and kept the ability to generate negative potential shifts and in the transfer
condition they continuously improved their ability to produce shifts in the positive direction A similar
observation could be made in the Theta/Beta group (feedback condition) where children continuously
improved their ability to produce shifts in the positive direction (i.e. enlarging the Theta/Beta ratio =
“deactivation”). [Strehl, Leins et al., 2006] already showed that children are able to regulate their SCPs
at the end of treatment, i.e. to differentiate between positivity and negativity tasks, and that this skill
remained stable for at least six month. At 2 year follow-up this significant difference between positivity
and negativity tasks in the SCP-amplitudes could no longer be observed. Yet it should be pointed out
that in the feedback condition this effect was caused by a decrease of the mean SCP amplitude in
positivity tasks, while the mean amplitude in negativity tasks did not change significantly (see Fig. 3).
As the skill to produce positive shifts does not seem to be as important in regard to abolishment of
symptoms, we are not expecting a negative influence on the clinical outcome. In the transfer condition
[Leins, Goth et al., 2007] had shown that at six month follow-up children were able to produce positive
SCP-shifts in positivity tasks for the first time since the beginning of treatment. At 2 year follow-up
this skill still is present. Though there is no significant difference between positivity and negativity
tasks in the transfer condition it can be seen in Fig. 3 that children were still able to produce the
required shifts, i.e. to produce positive shifts in positivity tasks and negative shifts in negativity tasks.
Participants in the Theta/Beta group were able to decrease the Theta/Beta ratio in activation
tasks ever since the beginning of treatment. The ability to increase the Theta/Beta ratio in deactivation
tasks improved continuously over time. At 2 year follow-up, patients are - as at the end of treatment -
still able to decrease the Theta/Beta ratio in activation tasks and increase the Theta/Beta ratio in
deactivation tasks. In transfer tasks, patients were not able to produce a positive Theta/Beta ratio at any
time of assessment, yet the difference between Theta/Beta ratio in activation and deactivation tasks has
nearly doubled from the end of treatment until 2 year follow-up.
Though mean SCP-amplitudes and mean Theta/Beta ratios indicate that children are still able to
self regulate cortical potentials, many results especially in the Theta/Beta group do not reach statistical
significance. This may be due to the small number of patients participating at the study. The decreasing
number of participants in long-term follow-up studies is a well-known problem.
Parents reported a reduction of behavioral problems. In each of the three questionnaires children
remained below the cut-off values. Some of these values have been reached for the first time. In the
230
SCP group effect sizes of these changes were mostly medium, while in the Theta/Beta Group small
effect sizes could be observed. Test results of attention testing also showed further improvements again
with medium effect sizes in the SCP group and small effect sizes in the Theta/Beta group (see Table 5
for a synopsis of all questionnaires and tests results).
Neither psychometric data nor neurofeedback data support a significant difference between both
training protocols. Being asked about their experience, many children from the SCP-group reported,
that in the first trials of each session it took them pretty long to find adequate strategies to control the
cursor. This subjective report is reflected by Figure 3 and Figure 4: While children in the SCP group
were not able to respond with appropriate shifts neither in positivity nor negativity tasks during the first
sessions, children in the Theta/Beta group were able to decrease the Theta/Beta ratio in activation tasks
from the beginning.
As can be seen from EEG training data improvements in attention and behavior do not reliably
match with the improvements regarding self regulation of brain activity in both groups. Especially in
the Theta/Beta group significant results in EEG data are very rare. Due to the absence of a control
group there is currently no proof that self-regulation training of frequency bands or SCPs leads to
improved behavior. In the preceding study [Strehl, Leins et al., 2006] had used Pearson's 2 test to
show a statistical correlation between self-regulation skills of SCPs and clinical outcome.
Unfortunately this analysis cannot be applied at this 2 year follow-up due to the small number of
participants.
How to explain the good clinical outcome given the low number of significant results in EEG
training data in the Theta /Beta group? The most obvious explanation is that non-specific factors might
contribute to the outcome. [Goth, 2006] determined predictors for clinical outcome and acquisition of
self-regulation skills in children undergoing treatment with training of SCPs or Theta / Beta bands.
While in the SCP group the acquisition of self-regulation skills at six months follow-up could be shown
to be a predictor for a good clinical outcome, factors not related to the neurofeedback training (e.g. IQ
and psychological wellbeing) were detected in Theta/Beta group.
Another important question cannot be answered by our data: 'Would the symptoms have
improved without any treatment, too? Would an improvement just be the normal course of the disease?'
Several studies already addressed this question. [Barkley, Fischer et al., 1990] showed that
approximately 80% of adolescents with a diagnosis of ADHD – despite treatment – still had significant
ADHD symptoms at 8 year follow-up. [Biederman, Mick et al., 2000] reported that a decline in
hyperactivity symptoms in untreated children with ADHD correlates with age, while inattentive
symptoms do not. Our study showed a continuous improvement in both hyperactivity and inattentive
symptoms, which indicates that there is an association between training of self regulation skills and
clinical outcome.
The heterogeneity of the samples in both groups regarding age, IQ, diagnosis and gender is a
major limitation of the study. Though both groups were matched for age and IQ, there are obvious
differences between both samples. Moreover biasing of the data in the Theta/Beta ratio group cannot be
ruled out, as children who participated in the 2 year follow-up had shown greater improvement in their
behaviour at preceding assessment points than children who did not participate at the 2 year follow-up.
231
In the NIMH MTA study Jensen et al showed that early advantages of medication treatment
compared with community care and behavior therapy were no longer present at 3 year follow-up.
Instead the treatment modalities applied in the study did not differ significantly in terms of reduction of
ADHD symptoms but all reached and remained below baseline levels at any point between 14 months
and 3 years. In contrast to the results of the NIMH MTA study, in our study children still improved
although treatment was terminated 2 years ago.
The present study has shown that children improved in behavior and attention after being treated
with neurofeedback and that those effects are stable or even improved two years after the last training
session took place. At 2 year follow-up approximately half of the children in both groups do no longer
meet the diagnostic criteria for ADHD at all. The long-term effects of neurofeedback can be considered
as a major advantage of this treatment compared with pharmacological treatment. Future studies
involving blind control groups should prove a causal relationship between neurofeedback treatment and
clinical outcome.
References
National Institutes of Health. Diagnosis and treatment of attention deficit hyperactivity disorder. Natl Inst Health Consens Dev
Conf Consens Statement. 1998:1 –37. Available at:
http://consensus.nih.gov/1998/1998AttentionDeficitHyperactivityDisorder110html.htm. Accessed August 24, 2006.
Barkley, RA, Fischer, M, Edelbrock, CS and Smallish, L, The adolescent outcome of hyperactive children diagnosed by research
criteria: I. An 8-year prospective follow-up study, J Am Acad Child Adolesc Psychiatry, 29(4), 546-57, 1990
Biederman, J, Mick, E and Faraone, SV, Age-dependent decline of symptoms of attention deficit hyperactivity disorder: impact
of remission definition and symptom type, Am J Psychiatry, 157(5), 816-8, 2000
Cohen, J, Statistical Power Analysis for the Behavioral Sciences., 2nd ed. Hillsdale, NJ, 1988
Conners, CK, Conners’ Rating Scale-Revised: Technical Manual in. Multi-Health Systems,North Tonawande, NY,1997
Drechsler, R, Straub, M, Doehnert, M, Heinrich, H, Steinhausen, HC and Brandeis, D, 1Controlled evaluation of a
neurofeedback training of slow cortical potentials in children with Attention Deficit/Hyperactivity Disorder (ADHD), Behav
Brain Funct, 3 35, 2007
DuPaul GJ, Barkley RA, Connor DF, Stimulants. in Attention-Deficit Hyperactivity Disorder: A Handbook for Diagnosis and
Treatment. Barkley, R. Guilford Press New York, NY,1998 510-551
Eyberg, S, Pincus D. Eyberg Child Behavior Inventory and Sutter-Eyberg Student Behavior Inventory-Revised, Eyberg Child
Behavior Inventory and Sutter-Eyberg Student Behavior Inventory-Revised in. Psychological Assessment Resources,Odessa,
FL,1999
Fuchs, T, Birbaumer, N, Lutzenberger, W, Gruzelier, JH and Kaiser, J, Neurofeedback treatment for attention-
deficit/hyperactivity disorder in children: a comparison with methylphenidate, Appl Psychophysiol Biofeedback, 28(1), 1-12,
2003
Goldman, L, Genel, M, Bezman, R and Slanetz, P, Diagnosis and treatment of attention-deficit/hyperactivity disorder in children
and adolescents. Council on Scientific Affairs, American Medical Association, Jama, 279(14), 1100-7, 1998
Goth, G, Neurofeedbacktherapy bei Kindern mit ADHS: Prädiktoren für den Erwerb kortikaler Selbstkontrolle und die klinische
Verbesserung, Unpublished doctoral Thesis, 65, 2006
Heinrich, H, Gevensleben, H, Freisleder, FJ, Moll, GH and Rothenberger, A, Training of slow cortical potentials in attention-
deficit/hyperactivity disorder: evidence for positive behavioral and neurophysiological effects, Biol Psychiatry, 55(7), 772-5,
2004
Jensen, PS, Arnold, LE, Swanson, JM, Vitiello, B, Abikoff, HB, Greenhill, LL, Hechtman, L, Hinshaw, SP, Pelham, WE, Wells,
KC, Conners, CK, Elliott, GR, Epstein, JN, Hoza, B, March, JS, Molina, BS, Newcorn, JH, Severe, JB, Wigal, T, Gibbons, RD
and Hur, K, 3-year follow-up of the NIMH MTA study, J Am Acad Child Adolesc Psychiatry, 46(8), 989-1002, 2007
232
Johnstone, SJ, Barry, RJ and Anderson, JW, Topographic distribution and developmental timecourse of auditory event-related
potentials in two subtypes of attention-deficit hyperactivity disorder, Int J Psychophysiol, 42(1), 73-94, 2001
Leins, U, Goth, G, Hinterberger, T, Klinger, C, Rumpf, N and Strehl, U, Neurofeedback for children with ADHD: a comparison
of SCP and Theta/Beta protocols, Appl Psychophysiol Biofeedback, 32(2), 73-88, 2007
Lubar, JF and Shouse, MN, EEG and behavioral changes in a hyperkinetic child concurrent with training of the sensorimotor
rhythm (SMR): a preliminary report, Biofeedback Self Regul, 1(3), 293-306, 1976
Lubar, JO and Lubar, JF, Electroencephalographic biofeedback of SMR and beta for treatment of attention deficit disorders in a
clinical setting, Biofeedback Self Regul, 9(1), 1-23, 1984
Monastra, VJ, Lubar, JF, Linden, M, VanDeusen, P, Green, G, Wing, W, Phillips, A and Fenger, TN, Assessing attention deficit
hyperactivity disorder via quantitative electroencephalography: an initial validation study, Neuropsychology, 13(3), 424-33, 1999
Monastra, VJ, Monastra, DM and George, S, The effects of stimulant therapy, EEG biofeedback, and parenting style on the
primary symptoms of attention-deficit/hyperactivity disorder, Appl Psychophysiol Biofeedback, 27(4), 231-49, 2002
Ramirez, PM, Desantis, D and Opler, LA, EEG biofeedback treatment of ADD. A viable alternative to traditional medical
intervention?, Ann N Y Acad Sci, 931 342-58, 2001
Ravens-Sieberer U, The KINDL Questionnaire for Measuring Health Related Quality of Life in Children and Adolescents-
Revised Version in Assessment of Quality of Life and Well-being. Schumacher J, KA, Brähler e. Hogrefe,Göttingen,
Germany,2003 184-188
Rockstroh, B, Elbert, T, Canavan, A, Lutzenberger, W and Birbaumer, N, Slow cortical potentials and behavior in. Urban &
Schwarzenberg,Baltimore,
München, Wien,1989
Rockstroh, B, Elbert, T, Lutzenberger, W and Birbaumer, N, Biofeedback: Evaluation and therapy in children with attentional
dysfunctions in Brain and behavior in child psychiatry. Rothenberger, A. Springer,Berlin,1990 345-355
Schachter, HM, Pham, B, King, J, Langford, S and Moher, D, How efficacious and safe is short-acting methylphenidate for the
treatment of attention-deficit disorder in children and adolescents? A meta-analysis, Cmaj, 165(11), 1475-88, 2001
Strehl, U, Leins, U, Goth, G, Klinger, C, Hinterberger, T and Birbaumer, N, Self-regulation of slow cortical potentials: a new
treatment for children with attention-deficit/hyperactivity disorder, Pediatrics, 118(5), e1530-40, 2006
Swanson, JM, Elliott, GR, Greenhill, LL, Wigal, T, Arnold, LE, Vitiello, B, Hechtman, L, Epstein, JN, Pelham, WE, Abikoff,
HB, Newcorn, JH, Molina, BS, Hinshaw, SP, Wells, KC, Hoza, B, Jensen, PS, Gibbons, RD, Hur, K, Stehli, A, Davies, M,
March, JS, Conners, CK, Caron, M and Volkow, ND, Effects of stimulant medication on growth rates across 3 years in the MTA
follow-up, J Am Acad Child Adolesc Psychiatry, 46(8), 1015-27, 2007
Tewes U, RP, Schallberger U, Hamburg Wechsler Intelligenztest für Kinder - Dritte Auflage (Hawik III). in. Huber,Bern,
Germany,1999
Zimmermann P, Flimm B, Testbatterie zur Aufmerksamkeitsprüfung (TAP) in. Psychologische Testsysteme,Herzogenrath,
Germany,2002
