A Sham-Controlled Study of Neurofeedback for Pain Management

BACKGROUND: Neurofeedback (NFB) attempts to alter the brain’s electrophysiological activity and has shown potential as a pain management technique. Existing studies, however, often lack appropriate control groups or fail to assess whether electrophysiological activity has been successfully regulated. The current study is a randomized controlled trial comparing changes in brain activity and pain during NFB with those of a sham-control group. METHODS: An experimental pain paradigm in healthy participants was used to provide optimal control of pain sensation. Twenty four healthy participants were blind randomized to receive either 10 × NFB (with real EEG feedback) or 10 × sham (with false EEG feedback) sessions during noxious cold stimulation. Prior to actual NFB training, training protocols were individually determined for each participant based on a comparison of an initial 32-channel qEEG assessment administered at both baseline and during an experimental pain task. Each individual protocol was based on the electrode site and frequency band that showed the greatest change in amplitude during pain, with alpha or theta up-regulation at various electrode sites (especially Pz) the most common protocols chosen. During the NFB sessions themselves, pain was assessed at multiple times during each session on a 0–10 rating scale, and ANOVA was used to examine changes in pain ratings and EEG amplitude both across and during sessions for both NFB and sham groups. RESULTS: For pain, ANOVA trend analysis found a significant general linear decrease in pain across the 10 sessions (p = 0.015). However, no significant main or interaction effects of group were observed suggesting decreases in pain occurred independently of NFB. For EEG, there was a significant During Session X Group interaction (p = 0.004), which indicated that EEG amplitude at the training site was significantly closer to the target amplitude for the NFB compared to the sham group during painful stimulation, but this was only the case at the beginning of the cold task. CONCLUSION: While these results must be interpreted within the context of an experimental pain model, they underline the importance of including an appropriate comparison group to avoid attributing naturally occurring changes to therapeutic effects.