This is a single group outcome, open label study in a clinical setting where both the patients and those administering treatment were aware of the treatment being given. Patients were given Informed Consent for biofeedback methods administered as well as Informed Consent to Research as put forth by the lead author's ethics committee of the American Psychological Association and the Association of Applied Psychophysiology and Biofeedback. All signed copies remain on file at The Better Brain Center. The authors had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Patients were recruited from the neurofeedback clinic of the Better Brain Center (formerly called Neurofeedback Consultants) where most were referred by their primary care physician or neurologist. Others responded to local media events covering their neurofeedback work in treating migraines. 74 headache clients presented to the clinic between 2004 and 2007. Selection criteria required that the client have migraine with or without aura and that this diagnosis be confirmed according to the IHS classification criteria for headache disorders [24
]. Patients who had less than one migraine per month or more than 20 per month were excluded from the study. This broad range of 1 to 20 migraines per month was used to keep as many patients as possible in the sample; however, we also conducted analyses using just the subset of patients with 2 to 14 migraines per month, the more typical headache frequency criterion used in migraine studies. The total sample included 37 migraine patients (29 females and 8 males). Ages ranged from 9 to 79, with the majority (56%) between the ages of 16 and 52, and the remainder evenly split between the younger group (22% were between 9 and 15) and the older group (22% were between 55 and 79). In terms of medical history, most patients had long, stable histories of migraine and had tried multiple pharmaceutical treatments prior to neurotherapy. All were having at least one migraine per month and taking at least one type of medication (preventive, abortive or rescue) for their migraines and were not required to discontinue these during the study (See Table ). About one-third of the patients had migraine with aura, and about three-fourths reported experiencing other kinds of headaches or one or more other significant conditions (e.g., anxiety, depression, problems with sleep or focus). All received a diagnosis of migraine or mixed migraine with tension type headache. Patients were screened for medication overuse headache and those taking abortive or rescue medications more than 2 or 3 times a week were referred to their prescribing physician for instructions on tapering down and possible alternatives to these types of medications.
Number of patients on each type of migraine medication
A personal and family headache history was taken at initial evaluation and a diagnostic interview was performed by a licensed psychologist to confirm the IHS-diagnosis of migraine with or without aura and to assess other symptoms and conditions. All patients had also received a diagnosis of migraine by a physician (neurologist, family practitioner or OB GYN) prior to entering this study. For patients who did not have at least two weeks of headache diaries, they were asked to wait two weeks to begin treatment in order to keep a baseline daily diary to record headache frequency and severity. At the first session and every 10 sessions thereafter, clients were asked to complete a non-standardized checklist to indicate changes in headaches as well as other symptoms (e.g., anxiety, insomnia, other pain types, depression, and behavioral problems). These checklists, daily headache diaries and clinical interviews were used throughout treatment to help determine most effective protocols and placements which were modified accordingly.
Follow up data collection
The data reported in this study were collected 3 months to 2 years after patients stopped coming to the neurotherapy center, either because they had completed the recommended number of treatment sessions or because they discontinued treatment on their own. The data were collected through follow-up telephone surveys conducted by a research consulting firm not affiliated with The Better Brain Center. Interviewers introduced themselves and indicated that they were conducting a follow-up assessment of The Better Brain Center's migraine patients for research purposes, and asked for permission to continue. Prior to starting treatment patients had been informed that their data may be used in a future retrospective study and that they might be interviewed for a study and all had agreed in writing. The telephone surveys were conducted during the last half of 2007 using a standardized protocol and prepared list of questions. The large majority of patients had completed treatment at least 6 months prior to the follow-up call, some as long as two years earlier. When asked about their post-treatment migraine history, participants were instructed to think about the 6 months immediately preceding the follow-up interview (not the entire time since their last treatment) and to estimate, on average, how many migraines they experienced per month. We included a broad range of months instead of just the 1 or 2 preceding months for two reasons. First, many patients experienced fewer than 1 per month post-treatment and a six month period is more likely to capture the fact that they are not completely migraine free. Second, many patients in this sample experienced fluctuations in the number of migraines they experienced per month, and capturing data for only 1 or 2 months could misrepresent their typical or average migraine frequency (although this is not a problem in large samples where unusually high or low frequency months should be randomly distributed across the sample; in smaller samples this is less likely to occur and short data collection periods can be an unnecessary source of error). Pre treatment headache frequency data was provided by headache diaries kept prior to treatment. Pretreatment headache frequency was also confirmed by interviews during follow up data collection. When asked about their migraine pattern prior to treatment, participants were also asked to recall the average number of migraines per month they were experiencing in the 6 months prior to seeking treatment. Interviewers had access to the migraine frequency data reported at the initial diagnostic evaluation, and if there were large discrepancies asked for clarification. Large discrepancies were rare, and were usually the result of misunderstanding the question or mixing headache types. In the 3 cases where participants could not be contacted or chose not to complete the telephone interview, migraine frequency data from patient records was used instead of the telephone survey data. In addition to the migraine frequency questions, we presented each respondent with a list of symptoms that are common among clients seeking neurotherapy (e.g., anxiety, focus or attention, depression, other (non-migraine) headaches, and asked which they had also experienced prior to treatment. For each symptom they reported having experienced we asked them to rate the level of improvement they experienced with treatment. The options for the 5 point rating scale included: 0 - No (0%) improvement, 1 Slight (10-30%) improvement, 2-Moderate (40-60%) improvement, 3-Major (70-90%) improvement, 4- Total (90-100%) improvement. We also asked all participants to use the same scale to rate the improvement they felt they had experienced in their migraines.
The study involved treatment using EEG biofeedback, pIR HEG biofeedback and handwarming biofeedback for an average total of 40 sessions. Average length of time in treatment was 6 months. Subjects underwent an average of 30 frequency-based neurofeedback sessions and 10 pIR HEG sessions for 30 minutes at least twice weekly. Eleven patients had an interruption in their treatment after the initial 20 sessions of up to several weeks but returned for their remaining sessions.
Assessment: A neurophysiological assessment using EEG measurements was administered using two channels of the EEG amplifier/software Brainmaster Atlantis version 3.5 or Brainmaster version 2.5. EEG Data was collected in three minute segments, 2 channels at a time. Data was collected under eyes open, eyes closed and eyes open task conditions at each site. Ten sites were collected: frontal (F3, FZ, F4), temporal T3, T4) central (C3, CZ and C4) and parietal areas (P3, P4) using the International 10-20 system of electrode placement.
Data collected was used to determine peak amplitudes for specific frequencies within the 1-38 HZ range. Data collection was performed at each site with the ground at FZ and with ipsilateral references and with the ground at C3 when sampling the FZ site. Sampling rate was 256 samples per second using a third order filter and a 60 HZ notch filter. Amplitude measures were peak to peak magnitude.
Treatment: The EEG measures were used to guide neurofeedback training protocols by targeting frequency ranges with the highest amplitude. All migraine patients were trained to reduce the amplitude of the targeted frequencies. The EEG training primarily occurred at 5 sets of homologous sites - (T3-T4, C3-C4, F3-F4, FP1-FP2 and P3-P4). These homologous sites were chosen according to the lead author's training in neurofeedback in which years of clinical experience in treating migraines by other experienced clinicians is taught [22
]. Electrode placements at homologous sites were used and training always began as a single channel placement using the first site as the signal and the second site as reference (example: T3-T4).
When training each site, each client received positive feedback (visual and auditory rewards via a video type of game) whenever EEG activity exceeded certain amplitude thresholds within a target band (called the reward band). Training each homologous site rewards the difference between the amplitude at each site of the target frequency. Simultaneously, clients received no feedback whenever excessive high amplitude activity occurred in certain frequency bands (called the inhibit band) that had been identified during the assessment. This lack of auditory reward (no beep or feedback) is designed to discourage the client from making excessive activity in the inhibit band whenever this activity exceeds the set threshold.
These protocols also reflect findings of EEG abnormalities commonly found in migraine patients [7
]. The reward and inhibit frequencies chosen for each site were also selected according to the individual's neurophysiological assessment and they were also based on a history of training recommendations considered to be optimal for migraine stabilization by the authors and other practitioners in the field [22
]. For example, for temporal lobe sites we rewarded a target range of 12-15 HZ. Lower target range rewards were selected at posterior and frontal regions. During training, sometimes reward frequencies were lowered by 2 HZ from the starting point. For T3-T4, for example, after starting at 12-15 HZ, we may move down to 10-13 HZ. Small changes in the targeted reward frequency can produce noticeable clinical improvement, readily identified by the migraine sufferer. For the migraine population, however, we have observed that lowering the target frequency further than 2 HZ frequently can have a destabilizing effect and may actually trigger a migraine.
Training Objectives: Patients were told simply to "make beeps" or an auditory reward which occurs when they achieve the training objectives of increasing the reward amplitude and reducing excessive amplitude in the targeted inhibit bands. The frequency ranges that were inhibited at all the sites ranged between 2 and 14 HZ (low inhibit) and between 15 and 38 HZ (high inhibit). The reward frequency range across all the sites was usually between 8-18 HZ, with a 3 HZ band width. Each set of sites had their own set of frequencies to be rewarded and inhibited and this stayed the same for each pair of sites for each individual patient.
Sessions commenced at the temporal locations (T3, T4) for three to five sessions and then moved to central areas (C3, C4), then frontal (F3, F4), prefrontal (FP1, FP2), and parietal (P3, P4) for typically one to two sessions at each location. Only these five pairs of sites were used with the migraine protocols. We find that starting at the temporal lobe locations seems to offer the most powerful relaxation and stabilization initially in order for us to be able to proceed to the other locations which may not always be as relaxing or stabilizing. If patients had an adverse reaction to any of these sites, we moved on to another location. On rare occasions, a patient would respond by becoming too activated by work at a certain site so we would move to other sites in the treatment regimen. A typical treatment regimen would resemble the following: Sessions 1-4- EEG biofeedback at the temporal lobes simultaneously with thermal handwarming biofeedback, session 5-9- EEG biofeedback at central, frontal and parietal areas simultaneously with thermal handwarming biofeedback, Session 10, 12, 14, 16, 18, 22, 26, 30, 34, 38- pIR HEG biofeedback, Sessions 11,13, 15,17, 19-21, 23-25, 27-29, 31-33, 35-37, 39 & 40- EEG biofeedback at various locations along the cortex (depending on patient presentation and response) simultaneously with thermal handwarming biofeedback. Although we do only one or two sites per session, we have found the best responses with migraineurs when we continue to move around the whole head at various placements because we find that this population can be easily irritated by the overtraining of one site.
For most patients 30 minutes of pIR HEG biofeedback was introduced at approximately their tenth visit. This involved the patients wearing a headset which is designed to be worn at FPZ (center of forehead) and watching a movie and being challenged to keep the movie playing as the reward threshold was re-set to higher temperatures. The movie would remain playing as long as the patient's forehead temperature increased as the threshold was re-set. After two sessions the frequency-based neurofeedback training was reintroduced, and generally conducted pIR HEG every 2nd or 3rd session along with neurofeedback for the remainder of the treatment period. Rationale for changing the order and number of each of these protocols was based on patient tolerance and effectiveness of each protocol and was always subject to change based on patient feedback. We have found that if we allow the patient to elevate their forehead temperature more than 2 degrees, or if they are worked too aggressively, it is more likely that this technique can have the side effect of causing rather than aborting a migraine. Therefore, patients were asked to gently but mindfully increase their temperatures in order to watch the movie and we did not challenge them aggressively to do so by too frequently resetting the reward threshold.
Thermal handwarming biofeedback was also used simultaneously along with the EEG biofeedback during clinic sessions. All patients were given thermal biofeedback units along with instructions for how to perform thermal biofeedback at home on the days they did not have a clinic session. An analysis of biofeedback in combination with home training was found to be more effective than therapies without home training [5
]. The objectives of this study are to show that the combination of both neurofeedback modalities along with thermal biofeedback and medication will significantly lessen the frequency of headaches experienced by the participants and that these effects will be more robust than thermal biofeedback-only approaches and more effective and enduring than traditional medication-only approaches.
Frequency-based EEG biofeedback protocols used the Neurocybernetics (EEG Spectrum, International, Canoga Park, CA) with Procomp amplifiers (Thought Technology, Montreal, Quebec, Canada) or the Brainmaster systems (Brainmaster Technologies, Oakwood Village, OH). Passive Infrared Hemoencephalography units were also used (Jeffrey Carmen, Manlius, NY). Thermal handwarming biofeedback utilized the SC-911 unit (Biomedical Instruments, Inc., Warren, MI).