Neurofeedback for Depression

Rationale: Neurofeedback (NF), non-invasive operant conditioning of neurophysiological brain activity within a brain-machine interface, has been claimed to represent a new and effective therapy for major depressive disorder (MDD). Although fundamental research data provide support for the potential validity of this approach, no sound scientific data substantiate therapeutic claims. Recently, we obtained pilot data that do support the presumed effectiveness of NF for MDD.

Objective: The objective of this proposal is to examine the efficacy of NF for MDD in a RCT.

Study design: Double-blind placebo controlled intervention study

Study population: 50 Depressed participants between 18-65 years

Intervention: Participants will be randomized (double blind) to active NF aimed to decrease alpha-asymmetry between left and right frontal areas or sham NF. The intervention will be carried out 3 times weekly over 6 weeks (total of maximum 18 NF sessions).

Main study parameters/endpoints: Primary outcome measures are clinical response and remission as assessed with severity measures. Secondary outcome measures are neurophysiological changes throughout the intervention.

To examine if neurofeedback (NF), by influencing prefrontal alfa-asymmetry, is an efficacious treatment for major depressive disorder.

Schematic overview of the study

Week 1 Week 2 Week 3 Week 4 Week 5 Week 6 FU
Baseline 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
SCID-1 interview x
QIDS x x x x x x x x x x x x x x x x x x x x
HDRS x x x x x x x
RDQ x x x x x x x x
STAI x x x
RAND36 x x x

All participants will be asked to fill out the QIDS, RDQ, and RAND-36, 3 months after termination of the intervention to assess stability of therapeutic effects. The HDRS and the QIDS-SR are conventional depression severity rating scales that are widely used in treatment studies for depression. The RDQ is a new instrument designed to assess clinical response and remission as well as improvement of functioning in daily life. In case participants start with a conventional treatment for depression before this 3 months follow-up period, they will be asked to fill out these questionnaires a few days prior to start of treatment. Additionally, a 5-minute baseline EEG will be measured to assess AA. These measurements will take place in our laboratory. Primary endpoints are scores on the QIDS-SR and HDRS as outlined in paragraph 8.1.1. Response and remission criteria for the QIDS-SR is defined as at least 50% reduction from the baseline score, and <6 respectively (see www.ids-qids.org). Response on the HDRS is defined as at least 50% reduction from the baseline score, remission is defined as a score <8 [24].

After signing informed consent, participants will be invited to the neurofeedback laboratory of the school for Mental Health and Neuroscience (MheNS). This laboratory is a facility that meets the requirements as outlined in NEN 60601 by the Nederlands Normalisatie-instituut. The intervention will take place while participants sit behind a table and look at a monitor on top of this table. The monitor displays a visual feedback signal (thermometer) based on real-time analysis of their electrophysiologidal data. The participants are instructed to raise the bar of the thermometer (which is indicative of a decrease in AA). During the NF sessions, while being in the laboratory, participants are in constant contact with the research assistant through an audio-channel, additionally there is constant video-surveillance. Prior to the NF session, depression-severity will be assessed as outlined above (see outcome measurements).

EEG-electrodes will be attached following the international 10-20 system at F3, F4, C3, C4, P3 en P4. EEG will be referenced by 2 electrodes attached to the earlobes. Under and above the left eye an electrode will be placed to measure EOG. The electrode will be applied after cleaning the skin with scrubgel. The impedance on all locations will be kept lower than 5 KΩ.

At the start of each NF session, the baseline EEG is being measures without NF to assess baseline AA which serves as a starting point for feedback. Measuring baseline prior to each session is necessary as can be expected that, at least in the active arm, AA decreases over time. Data collection will be channeled through an acquisition PC with a BrainAmp DC EEG amplifier (Brain Products) using a 1000 Hz sample frequency. Online calculations are done by a filter written for BrainVision RecView. The data will be epoched online into 2.048-s epochs that overlap by 75% and then transformed by a fast Fourier transform (FFT) to the frequency domain (frequency resolution 0.488 Hz). Every 0.512 second, the power within the alpha frequency band (7.8 Hz – 13.1 Hz) of both F3 and F4 will be calculated. AA is computed as the difference of the natural log-transformed F3 and F4-alpha power: Ln(F3-alpha) – Ln(F4-Alpha). Current asymmetry is subsequently compared to the personal mean baseline asymmetry. The result of the calculation will be sent to a stimulus PC running Presentation stimulus delivery software (Neurobehavioral Systems) with an 8-bit parallel port (LPT-port) to control a paradigm showing a visual representation of the asymmetry. In the Presentation paradigm, the last 20 values of the asymmetry are used in a moving average to prevent ‘jitter’ in the feedback. Participants receive feedback with visual feedback; they are instructed to increase the level of a thermometer that is shown on a flatscreen. Additionally, a numerical score below the thermometer indicates their actual total performance. This score is adjusted (i.e. increased) continuously by a number ranging from 0 and 128, depending on the level of the thermometer. In this way a good actual performance (a shift in asymmetry in the desired direction) results in an increasing total score. A big shift in the desired direction results in a rapidly increasing total score, whereas a small shift in the desired direction results in a slow increasing total score. A shift in the undesired direction produced no change in total score. The purpose of this total performance score is to give participants feedback on the differential effect of the sessions.

The sham-NF will be delivered with the same data acquisition protocol, but the AA will be randomly multiplied by +1 or -1, which reduces the correlation between measured brain activity and the visual feedback signal to almost zero. Pilot-data showed that healthy subjects were not able to distinguish between real and sham-NF.



A maximum of 18 NF sessions will be given during 6 weeks. The intervention will be stopped if participants meet criteria for remission (both QIDS-SR and HDRS) before they have received this maximum of 18 sessions. If no clinical response occurs, participants will be offered an evidence-based treatment at the unit for treatment of mood disorders at the RIAGG Maastricht as planned.