102 children with ADHD (8–12 years) participated in a NF training or an AST. Subjects were randomly assigned to one of the two study groups (ratio NF, control training = 3:2; see also Fig. 2). Eight children (NF, n = 5; AST, n = 3) discontinued the study due to immediate need for medical treatment (n = 3), organisational problems of the parents (n = 2), loss of motivation (n = 1) or protocol violation (n = 2). Sample size had been estimated a priori to be large enough to detect a medium effect size of about 0.5 with a power of 0.8 (one-sided, 0.05-level test).
Table 1 summarises inter alia demographic, psychological and clinical variables of the children completing their training. Concerning these variables, there were no significant differences between NF group (n = 59) and control group (n = 35).Footnote 1
All patients fulfilled DSM-IV criteria for ADHD . Diagnoses were based on a semi-structured clinical interview (CASCAP-D ) and confirmed using the Diagnostic Checklist for Hyperkinetic Disorders/ADHD  by a child and adolescent psychiatrist or a clinical psychologist, supervised by a board-certified child and adolescent psychiatrist. With mean total FBB-HKS scores of about 1.5 (range 0–3; see Table 2), ADHD symptomatology was moderately pronounced in both training groups. Children with comorbid disorders other than conduct disorder, emotional disorders, tic disorder and dyslexia were excluded from the study. All children lacked gross neurological or other organic disorders. All children were drug-free for at least 6 weeks before starting the training and without concurring psychotherapy.
The study follows the CONSORT guidelines for randomised trials . It was approved by the local ethics committees of the participating clinics and conducted according to the declaration of Helsinki. Assent was obtained from the children and written informed consent from their parents.
Design of the study
The design of the study is illustrated in Fig. 1. NF and AST both consisted of 36 units of 50 min each. Both treatments were divided in two blocks of 18 units. These 18 units were combined in 9 sessions. These sessions took place two to three times a week. The NF training consisted of 1 block of 18 units of theta/beta training and 1 block of 18 units of SCP training (balanced order). For both NF and AST training, there was a break of 2–3 weeks between the two treatment blocks. The NF and the AST training were designed as similarly as possible concerning the setting and the demands upon the participants, e.g., both treatments encompassed attention demanding tasks on a computer (to a comparable amount, 25–30 min per training unit), acquirement of strategies for focussing attention, and efforts to transfer learned strategies into daily life . Parents were not explicitly informed about the treatment condition of their child (NF vs. AST).Footnote 2
Parent ratings were assessed in the week before the training course started (pre-training), about 1 week after the last session of the first (intermediate
Footnote 3) and second training block (post-training), respectively, and 6 months after the end of the training (follow-up).
In contrast to the other assessment points, neuropsychological/physiological data were not measured at follow-up.
The NF system SAM (“self-regulation and attention management”) which was developed by our study group was used for NF training.
In theta/beta training, the task was to reduce theta and enhance beta activity. A bar on the left of the screen (representing theta activity) had to be reduced while simultaneously a bar on the right (representing beta activity) had to be increased. In each unit, about five or six trials of 5 min each, or up to three trials of 10 min each, were performed. Baseline values of theta and beta activity were determined at the beginning of each session (3 min). Children were instructed to reach a relaxed but attentive state and to find individual strategies to control the bars.
In SCP training, the children had to generate negative or positive SCPs. They had to find appropriate strategies to direct a ball upwards (negativity trials) or downwards (positivity trials). Negativity (50%) and positivity trials (50%) were presented in random order. A trial lasted for 8 s (baseline period: 2 s, feedback period: 6 s). Children were instructed to reach an attentive (negativity trials) or relaxed state (positivity trials). In each SCP training unit, approximately 120 trials were performed, divided into 2–3 blocks of 40–60 trials in each treatment unit.
For both NF protocols, feedback was calculated from Cz (reference: mastoids, bandwidth: 1–30 Hz for theta/beta training and 0.01–30 Hz for SCP training, respectively, sampling rate: 250 Hz). Vertical eye movements, which were recorded with electrodes above and below the left eye, were corrected online using slightly different regression-based algorithms for theta/beta training  and SCP training . For segments containing artefacts exceeding ±100 μV in the EEG channel and ±200 μV in the EOG channel, no feedback was calculated.
Transfer trials, i.e., trials without contingent feedback, were conducted about one-third at the beginning of a training block and about two-third at the end of a training block. These transfer trials, as well as homework, were intended to improve generalisation of treatment effects.
Attention skills training
The AST was based on “Skillies” (Auer-Verlag, Donauwörth, Germany), a German learning software programme which primarily exercises visual and auditory perception, vigilance, sustained attention, and reactivity. In “Skillies”, the children had to sail to several islands. On each island, a clearly defined task—each requiring different attention-based skills—had to be solved (for further information, see ). The training was complemented by some self-directed interventions from cognitive therapy to assure comparability to NF, i.e., the children were to compile (meta-)cognitive strategies such as focusing attention, careful processing of tasks and impulse control. Corresponding to the NF group, children of the AST group should practice their compiled strategies in daily life situations.
The following questionnaires (assessed at pre-training, post-training and follow-up) were completed by parents to evaluate the follow-up results:
German ADHD rating scale (FBB-HKS) : The FBB-HKS is a 20-item questionnaire related to DSM-IV and ICD-10 criteria for ADHD and hyperkinetic disorders, frequently used in Germany in the evaluation of medical and cognitive behavioural treatment of ADHD (e.g. ). The severity of each item is rated from 0 to 3. Outcome measures were the main FBB-HKS total score, i.e., the mean value of all items as well as subscores for inattention and hyperactivity/impulsivity. The FBB-HKS total score constituted the primary outcome measure of the study.
German Rating Scale for Oppositional Defiant/Conduct Disorders (FBB-SSV) : It is comprised of 25 items. The severity of each item is rated from 0 to 3. Outcome measures were the subscales oppositional behaviour (mean value of the first 9 items) and delinquent and physical aggression (mean value of the remaining 16 items).
The Strength and Difficulties Questionnaire (SDQ, German version) [24, 33] is comprised of 25 items which address both positive and negative attributes. Each item is rated from 0 to 2. Outcome measures were the total difficulties score as well as the five subscales (emotional symptoms, conduct problems, hyperactivity, peer problems, prosocial behaviour).
The Home Situations Questionnaire (HSQ, German version)  was used to assess behaviour problems of the child in specific home situations. The HSQ consists of 16 situations in which problematic child behaviour can occur. Parents rate whether the problem behaviour is present in that setting; if so, they rate its severity on a nine-point scale.
Problem behaviour during homework was assessed using the Homework Problem Checklist (HPC, German version) . This checklist consists of 20 items, rated on a four-point frequency scale.
Per-protocol analysis was conducted to avoid confounding the treatment effects with additional treatment. Children were classified as dropouts and excluded if they had started another treatment (e.g., medication, psychotherapy) or if questionnaires were not returned.
Behavioural data were analysed in repeated-measures ANOVAs with between-subject factor GROUP (NF vs. control training), within-subject factor TIME (post-training, follow-up) using the baseline (pre-training) measure as a covariate. If NF training effects are still superior to the control training at follow-up, the ANOVA is expected to reveal a significant GROUP effect. If the difference between the two trainings becomes significantly greater or smaller, this effect is expected to be indicated by a significant GROUP × TIME interaction.
Effect sizes (Cohen’s d) were calculated as the difference of the change of a measure from pre-training to an assessment point (post-training and follow-up, respectively) and the corresponding change score in the control group divided by the pooled standard deviations of these change scores. To compare the ratio of responders (≥25% reduction of the primary outcome measure) in the NF group and the control group, the odds ratio was calculated.
Though limited by a smaller sample size, possible effects of the order of the NF training blocks were tested by comparing improvements in the FBB-HKS obtained at post-training and at follow-up (t tests).
Since there were about 35% dropouts (see “Results”), we also tested for differences in the clinical (behavioural) data between the dropouts and the remaining children. We computed ANOVAs with between-subject factors GROUP (NF vs. control training) and DROPOUT (dropout vs. follow-up data available) and within-subject factor TIME (pre-training, post-training).
SPSS (v.16) was used for statistical analysis. For all statistical procedures, significance was assumed if p < 0.05.