Abstract
Objective:
Mindfulness-based interventions (MBI) have been shown to reduce subjective symptoms of insomnia but the effects on objective measures remain unclear. The purpose of this study was to examine sleep EEG microarchitecture patterns from a randomized controlled trial of Mindfulness-Based Stress Reduction (MBSR) and Mindfulness-Based Therapy for Insomnia (MBTI).
Methods:
Sleep EEG spectral analysis was conducted on 22 participants with chronic insomnia (>6 months) randomized to 8-week MBSR, MBTI, or self-monitoring control (SM). Overnight polysomnography with 6-channel EEG was conducted at baseline, post-treatment, and 6-month follow-up. Spectral power averaged from channels C3/C4 across NREM epochs (excluding N1) was examined for within-group changes and relationships with self-report measures.
Results:
Increases in absolute NREM beta (16–25Hz) power were observed from baseline to post-treatment (p=0.022, d=0.53) and maintained at 6-month follow-up (p=0.011, d=0.57) in the combined MBI groups, and additionally in the gamma (25–40Hz) range at follow-up for the MBTI group only. No significant changes in these frequency bands were observed for SM. Following mindfulness intervention, NREM beta was positively associated with Five-Facet Mindfulness (FFM) score (rho=0.37, p=0.091) and Insomnia Severity Index (rho=−0.43, p=0.047).
Conclusion:
These results in people with insomnia corroborate prior reports of increased high-frequency sleep EEG power associated with mindfulness training. This change in beta EEG pattern merits further evaluation as a potential marker of the effects of mindfulness meditation on sleep, especially given the contradictory findings in the context of insomnia.
Clinical Trial Registration:
Keywords: mindfulness, meditation, sleep, insomnia, arousal, EEG
INTRODUCTION
The practice of mindfulness meditation has been shown to have many health benefits [1–3]. One area that has received burgeoning interest is the use of mindfulness-based interventions (MBIs) for insomnia, a condition that is a pervasive public health problem[4]. Approximately a third of the general population reports experiencing insomnia symptoms and approximately 10–18% meet criteria for an insomnia disorder, characterized by difficulties with sleep initiation or maintenance, or non-restorative sleep, and accompanied by significant impairments to daytime functioning[5]. Several randomized controlled trials using MBIs for insomnia have now been conducted, which collectively reveal large effect sizes on self-reported global measures of insomnia but the effects on objective measures of sleep are less clear[6–8].
Models of insomnia posit that hyperarousal is a prominent pathophysiological mechanism in the progression of chronic insomnia[9,10], and many studies have reported increased high-frequency EEG activity (>15Hz) as an indicator of cortical hyperarousal[11–15]. Most treatments for insomnia have historically aimed to decrease arousal and correspondingly, anticipate this beneficial decrease in arousal to be associated with an increase in slow wave sleep (SWS) or slow wave activity (SWA) and decrease in high-frequency EEG[16].
Mindfulness practices are complex and multifaceted, with the goal of increasing awareness and insight. Mindfulness practices support metacognitive shifts, such as observing pre-sleep worry rather than actively attempting to change the distressing thoughts and any associated physical sensations. Consequently, an increase in both relaxation and alertness in EEG patterns from meditation training has been hypothesized by Britton et al.[17] Ferrarelli et al.[18] found evidence to support this hypothesis in healthy, long-term meditation practitioners with extensive training (1,500–30,000+ hours of lifetime practice). Compared to participants with no prior meditation experience, these experienced practitioners exhibited significantly greater NREM EEG activity in the gamma (25–40Hz) frequency range, posited to reflect cognitive arousal, over a parietal-occipital region of the scalp and directly correlated with the hours of lifetime practice. The authors postulated that the increased gamma activity is due to sustained, plastic changes to neural circuits involved with meditation practice that are observed during NREM sleep as a window into trait changes of brain functioning. The waking EEG literature has concurrently reported similar findings of meditation-related increases in high-frequency (>15Hz) EEG activity thought to reflect increased alertness[19,20].
The observation of increased high-frequency EEG activity associated with meditation practice, alongside self-reported improvements in sleep, presents a potential paradox in the context of insomnia. As noted above, high-frequency EEG activity is one of the most commonly reported physiological correlates of insomnia and is thought to be an intrinsic part of maladaptive hyperarousal. To our knowledge, only one study has examined this paradox. Britton et al.[21] conducted polysomnography recordings in conjunction with a Mindfulness-Based Cognitive Therapy (MBCT) intervention for individuals with recurrent history of depression and residual sleep disturbance. An increase in stage 1 minutes and decrease in SWS was observed following MBCT, each proportional to the amount of meditation practice. Furthermore, post-hoc spectral analysis of the EEG data found an increase in NREM gamma range (25–40Hz) activity[22]. Although these findings may appear to reflect a maladaptive increase in light sleep and decrease in deep sleep, they were directly concurrent with improvements in multiple self-report measures of sleep quality[21,22], consistent with the notion of ‘relaxed awakening’[17].
The present study was a secondary analysis examining microarchitecture of sleep EEG from a randomized controlled trial comparing Mindfulness-Based Stress Reduction (MBSR) and Mindfulness-Based Therapy for Insomnia (MBTI) with a self-monitoring (SM) control for people with insomnia. Previous reports from this trial have focused on the clinical outcomes for insomnia[23], and self-reported cognitive-emotional measures[24]. The aim of this study was to assess both short and longer-term changes in EEG activity among participants who received an MBI, focusing on the high-frequency beta (16–25Hz) and gamma (25–40Hz) ranges.
METHODS
Participants
Participants for this study were from a single-site randomized controlled trial conducted at Rush University Medical Center between November 2008 and February 2012. Details on recruitment and screening procedures for the parent trial have been previously described[23]. All participants were adults over the age of 21 who met criteria for insomnia disorder[25] (difficulty initiating or maintaining sleep despite adequate opportunity for sleep, with at least one symptom of an associated daytime impairment) with self-reported sleep onset latency (SOL) or wake after sleep onset (WASO) >30 minutes at least 3 nights per week, for at least the past 6 months. In addition, participants reported at least one symptom of heightened cognitive or somatic arousal (e.g. anxiety about sleep, elevated muscle tension) based on ICSD-2 criteria for psychophysiological insomnia. Exclusion criteria for the study included uncontrolled medical or psychiatric condition that could interfere with sleep or required immediate treatment, comorbid sleep disorders screened using polysomnography (e.g., sleep-disordered breathing, periodic limb movements), or the use of sedative hypnotics.
A total of 54 participants met study criteria and were randomized to one of three conditions in the parent study (see Ong et al.[23,24] for details) with 19 randomized to MBSR, 19 randomized to MBTI, and 16 randomized to the SM control. Overnight polysomnography (PSG) was assessed at baseline, post-treatment (8 weeks), and 6-month follow-up for the MBSR and MBTI groups and at baseline and post-monitoring (8 weeks) for the SM group. The SM group was offered behavior therapy after the post-monitoring assessment and therefore did not have a naturalistic follow-up. For the present study, analyses were confined to participants with complete EEG data for their respective groups. The study protocol was approved by the Rush University Medical Center institutional review board, registered on clinicaltrials.gov (NCT00768781), and all participants provided written informed consent during the screening interview.
Treatment Arms
Mindfulness-based stress reduction (MBSR)
The MBSR treatment in this study followed the standard protocol used at the Center for Mindfulness at the University of Massachusetts Medical School. The program consisted of 8 weekly group meetings lasting 2.5 hours each plus one 6-hour meditation retreat held between the 5th and 7th week. Each group meeting included meditation practice (breathing meditation, body scan meditations, walking meditations, Hatha Yoga), a period of general discussion about the at-home meditation practice, and education on the daily applications of meditation. MBSR was taught by 2 instructors with doctoral degrees (PhD or MD) who had over 2 years of experience teaching MBSR. Neither had formal training in sleep medicine or CBT for insomnia.
Mindfulness-based therapy for insomnia (MBTI)
The MBTI treatment consisted of 8 weekly group meetings lasting 2.5 hours each plus one 6-hour meditation retreat held between the 5th and 7th week. Each group meeting session began with formal mindfulness meditations that include one quiet (body scan, breathing, sitting meditation) and one movement meditation (yoga, walking, stretching meditation). A period of discussion led by the MBTI instructor followed, however unlike the general discussion about the at-home meditation practice and education on the daily applications of meditation in the MBSR treatment, participants in MBTI discussed the application of mindfulness principles to the problem of insomnia. MBTI provided specific behavioral strategies for insomnia (sleep restriction therapy, stimulus control, and sleep hygiene) delivered within the context of mindfulness principles. MBTI was delivered by the senior author, who has specialized training in mindfulness meditation and behavioral treatments for insomnia.
In both MBSR and MBTI, participants instructed to practice meditation at home for 30–45 minutes at least 6 days/week and were asked to keep a meditation diary along with their sleep diary. In addition, they were provided with the book Full Catastrophe Living by Kabat-Zinn[26] and a CD for guided meditation to aid in their home practice.
Self-Monitoring Control (SM)
SM consisted of an 8-week period during which participants were instructed to complete sleep/wake diaries daily to control for the effects of self-monitoring by completing diaries, which was also used in MBSR an MBTI. Upon randomization to SM, participants attended an orientation session during which the purpose and instructions for the sleep dairies were provided along with the materials for completing the diaries. To enhance motivation, participants were told that the diary data completed during the SM would be used to inform treatment planning during the behavior therapy that followed the SM period. SM was managed by a member of the research team, who would contact participants if diaries were not sent in on time or if participants had questions but otherwise had no structured contact with participants.
Polysomnography (PSG) and EEG spectral analysis
Laboratory-based, technician-monitored PSG was conducted for one night at baseline, post-treatment, and 6-month follow-up. For the baseline PSG, a standard PSG was conducted following standard practices[27], using 19 channels including EEG, EOG, chin and bilateral anterior tibialis EMG, EKG, snoring, airflow, respiratory effort, and pulse oximetry. For post and 6-month follow-up, a reduced montage without respiratory and leg EMG measures were used. For all PSGs, scoring of sleep parameters followed American Academy of Sleep Medicine standards[28] and were performed by research staff under the supervision of a registered polysomnography technologist (RPSGT). Bedtimes and wake times during the PSG were based upon habitual sleep/wake patterns derived from the screening/baseline diary with a target of 8 h as the time-in-bed (TIB) interval (i.e., lights off to lights on). At post and 6-month follow-up, participants who completed MBTI were allowed to reduce TIB based upon recommendations received during the intervention.
EEG data from the available six channels (F3, F4, C3, C4, O1, and O2), sampled at 256Hz, were extracted and processed in MATLAB (Mathworks, Nattick, MA). Signals were referenced to the contralateral mastoid and filtered using a 0.5–105Hz bandpass then a 55–65Hz notch filter to remove 60Hz noise while still enabling analysis of higher frequencies. Similar to prior studies[29,30], a semi-automatic artifact rejection and data cleaning procedure was utilized (performed blind to study group and timepoint). Specifically, power in low (1–4Hz) and high (16–40Hz) frequency bands across all epochs for each channel of each recording were reviewed, and notably high powered-epochs were removed, which often corresponded to the 99.5th percentile of power distribution for that channel and recording. Exceptions to this guideline were made depending on visual salience of the power distribution during inspection. This data cleaning procedure was performed blind to study group and timepoint, thus minimizing the risk of influence from movement and other artifact particularly in the high frequency range of interest. Spectral analysis of the cleaned 6-second NREM sleep epochs (excluding N1) was performed via Welch’s averaged modified periodogram with Hamming window using the function ‘pwelch’ in MATLAB. Given that the few relevant existing reports of sleep EEG changes associated with mindfulness have found effects most prominently in central channels, the average of C3 and C4 was used as the primary EEG measure. Data were analyzed from 1–100Hz due to the exploratory purpose of this report. EEG frequency bands were defined based on commonly used cutoffs: delta (1–4Hz), theta (4–8Hz), alpha (8–12Hz), sigma (12–16Hz), beta (16–25Hz), gamma (25–40Hz), omega (40–100Hz). Given the interest in prior studies of insomnia to evaluate relative EEG power across standard frequency bands (each band divided by total power), these secondary comparisons were also conducted using averaged C3/C4 NREM data.
Self-report measures
The Insomnia Severity Index (ISI)[31] is a validated 7-item scale (range = 0–28) that assesses the severity of both nighttime and daytime symptoms of insomnia over the past week.
The Pre-Sleep Arousal Scale (PSAS)[32] is a 16-item (range = 16–80) measure assessing somatic and cognitive arousal in the period prior to sleep. The PSAS total score may be used as a waking correlate of insomnia in this sample given that all participants reported symptoms of psychophysiological arousal.
The Glasgow Sleep Effort Scale (GSES)[33] is a 7-item (range = 0–14) measure of sleep effort during the past week. In the present study, the GSES was given as a state measure of arousal associated with sleep effort.
The Hyperarousal Scale (HAS)[34] is a 26-item (range = 26–104) measure of daytime arousal among people with insomnia. Higher scores reflecting elevations in arousal during wakefulness, which has been identified as a key characteristic of individuals with chronic insomnia.
The Five Facet Mindfulness (FFM)[35] questionnaire is a 39-item questionnaire that measures five facets of mindfulness training (awareness; labeling experience; nonjudgment; nonreactivity; and acting with awareness) found both in beginning meditation training and in clinical interventions.
The measures above were collected at baseline and post-treatment for all groups, and also at the 6-month follow-up for the MBSR and MBTI groups. In addition, the MBSR and MBTI groups completed prospective daily meditation diaries on the frequency and duration of meditation activity following the second session when meditations were introduced. Participants were asked to keep track of their daily meditation practice including the type of meditation they practiced and number of minutes they practiced.
Data Analysis
The data analysis plan consisted of three sets of analyses. The first set of analyses examined short-term changes in spectral EEG measures from baseline to post-treatment/monitoring using linear mixed models using SAS PROC MIXED (SAS Institute Inc., Cary, NC) with repeated observations nested within participants, time as a repeated categorical predictor, random intercept with variance components structure, and restricted maximum likelihood estimation. Given that the aims focused on beta and gamma EEG for those who received an MBI, we constructed separate models for the combined MBI groups and the SM group with absolute beta and gamma EEG as the dependent variables. These main analyses were followed by exploratory analyses using the same models on the absolute power of the other EEG bands (delta, theta, alpha, sigma, omega) and the relative power of all EEG bands as dependent variables for comprehensive purposes. Uncorrected post-hoc comparisons were conducted for MBSR and MBTI separately. The second set of analyses examined long-term changes in spectral EEG measures from baseline to 6-month follow-up. Since the SM group did not have follow-up data, these analyses were confined to the MBI groups using a similar approach to the first set of analyses except the time parameter included baseline, post, and follow-up. The third set of analyses examined the association between EEG variables evidencing significant change over time and self-reported variables to provide context for the EEG findings. Correlations (Spearman’s rho) were computed for each time point between the EEG measure and each of the self-report measures for the two MBI groups.
No a priori power analysis was conducted for these analyses, since they were secondary analyses (power considerations for the overall randomized controlled trial was based on the main clinical outcomes). Because of the exploratory, hypothesis-generating aim of this study, results significant at α=0.10 are reported, given the need to consider both Type I and Type II errors in hypothesis generating studies. Effect sizes (Cohen’s d) are provided as a further indication of the magnitude of change.
RESULTS
Participant characteristics
Thirty-six participants with complete data across the three timepoints were included in the present analyses (MBSR = 9; MBTI = 13; SM = 14). See Figure 1 for CONSORT flow diagram and reasons for incomplete data. The mean age of participants was 43.97 (SD=10.76, range=25–65) and 75.0% of the sample identified as female, 16.7% Hispanic/Latino, 63.9% Caucasian, 25.0% African American, 5.6% Asian, and 5.6% other.
Figure 1.
CONSORT flow diagram
EEG measures
Data cleaning
EEG data cleaning was comparable across timepoints with 91.5% of epochs and 97.5% of channels overall retained for analysis (main effect of Time: F(2, 56)=0.62, p=0.5407, and F(2, 56)=0.54, p=0.5853, respectively). Resulting individual power spectra demonstrated similar variability across time and mindfulness groups (Supplementary Figure 1). Data for one participant were winsorized for these analyses and subsequent correlations due to the values exceeding 3 standard deviations from the mean at all three timepoints.
Short-term Effects
See Table 1 for mean values and standard deviations for absolute power of averaged C3/C4 NREM EEG power across EEG frequency bands at each time point. For the MBI groups, linear mixed models demonstrated a significant increase from baseline to post-treatment for beta EEG (16–25Hz), F(1, 21)=6.11, p=0.0221, d=0.53. Post-hoc tests of each MBI group revealed a significant increase in the MBSR group, F(1,8)=4.69, p=0.0622, d=0.72, but no significant difference in the MBTI group, p>0.10. No significant results were found for the MBI groups on gamma EEG (25–40Hz) and no significant results were found for the SM group on either beta or gamma EEG. When the combined MBI group was compared to SM, no significant difference was found in the change in beta or gamma EEG over time.
Table 1.
Means and standard deviations of average absolute C3/C4 NREMEEG at each timepoint.
| Baseline | Post | Follow-up | |
|---|---|---|---|
| M (SD) | M (SD) | M (SD) | |
| Delta (1–4Hz) | |||
| MM | 74.30 (33.47) | 76.37 (33.82) | 70.93 (30.54) |
| MBTI | 68.50 (31.75) | 68.86 (30.54) | 62.32 (24.52) |
| MBSR | 82.68 (35.98) | 87.21 (37.13) | 83.37 (35.38) |
| SM | 71.75 (50.19) | 61.73 (39.45) | |
| Theta (4–8Hz) | |||
| MM | 9.95 (4.17) | 10.66 (4.58) | 10.52 (4.62) |
| MBTI | 9.54 (4.54) | 10.05 (4.85) | 9.95 (4.99) |
| MBSR | 10.53 (3.75) | 11.53 (4.28) | 11.34 (4.17) |
| SM | 10.22 (5.72) | 9.94 (5.92) | |
| Alpha (8–12Hz) | |||
| MM | 5.49 (3.53) | 5.73 (3.30) | 5.72 (3.30) |
| MBTI | 6.46 (4.27) | 6.42 (3.90) | 6.48 (4.02) |
| MBSR | 4.08 (1.30) | 4.74 (1.99) | 4.63 (1.40) |
| SM | 5.74 (3.80) | 5.67 (4.12) | |
| Sigma (12–16Hz) | |||
| MM | 2.97 (2.03) | 2.87 (1.40) | 3.34 (2.21) |
| MBTI | 3.54 (2.29) | 3.14 (1.58) | 3.92 (2.62) |
| MBSR | 2.14 (1.26) | 2.49 (1.06) | 2.51 (1.09) |
| SM | 2.57 (1.60) | 2.66 (2.00) | |
| Beta (16–25Hz) | |||
| MM | 0.42 (0.17) | 0.46 (0.20) | 0.50 (0.23) |
| MBTI | 0.47 (0.18) | 0.51 (0.23) | 0.56 (0.28) |
| MBSR | 0.35 (0.14) | 0.39 (0.13) | 0.41 (0.10) |
| SM | 0.42 (0.19) | 0.44 (0.21) | |
| Gamma (25–40Hz) | |||
| MM | 0.12 (0.08) | 0.13 (0.07) | 0.14 (0.08) |
| MBTI | 0.13 (0.08) | 0.15 (0.08) | 0.16 (0.10) |
| MBSR | 0.11 (0.07) | 0.10 (0.05) | 0.11 (0.03) |
| SM | 0.12 (0.08) | 0.12 (0.06) | |
| Omega (40–100Hz) | |||
| MM | 0.04 (0.03) | 0.05 (0.04) | 0.05 (0.03) |
| MBTI | 0.04 (0.02) | 0.05 (0.04) | 0.04 (0.03) |
| MBSR | 0.04 (0.04) | 0.04 (0.03) | 0.05 (0.02) |
| SM | 0.04 (0.03) | 0.04 (0.03) | |
MM: Mindfulness Meditation; MBTI: Mindfulness Based Therapy for Insomnia, MBSR: Mindfulness Based Stress Reduction; SM: Self-Monitoring.
Exploratory analyses conducted on the absolute power of other EEG bands revealed a significant increase from baseline to post-treatment for the MBI groups on omega EEG (40–100Hz), F(1, 21)=3.13, p=0.0914, d=0.38. Post-hoc analysis revealed a significant increase in the MBTI group, F(1,12)=3.75, p =0.0768, d=0.54, but no significant change in the MBSR group. There was a significant decrease from baseline to post-monitoring for the SM group on delta EEG (1–4Hz), F(1, 13)=4.69, p=0.0495, d=0.58. No other significant results were found. Linear mixed models conducted on relative power revealed no significant results (see Supplementary Table 1).
Long-term Effects
Linear mixed models conducted on the MBI groups from baseline to 6-month follow-up revealed a significant increase in absolute power for beta EEG, F(2, 42)=5.04, p=0.0110, d=0.68. Post-hoc analyses showed a significant increase in the MBTI group, F(2, 24)=5.22, p=0.0131, but not the MBSR group. Within the MBTI group, post-hoc tests showed a significant increase relative to baseline at follow-up (t[24]=−3.23, p=0.0036, d=0.77) and a significant increase from post-treatment to follow-up (t[24]=0.0973, p=0.0973, d=0.54). There was no significant change in gamma EEG power over time for the MBI groups combined. Post-hoc analyses revealed a significant increase in gamma within the MBTI group from baseline to follow-up (t[12]=−2.17, p=.0505, d=0.60), but not for the MBSR group.
Exploratory analyses conducted on absolute power of other EEG bands revealed a significant increase from baseline to follow-up in sigma EEG (12–16Hz), F(2, 42)=2.48, p=0.0962. Post-hoc analysis by group showed that this significant increase only occurred in the MBTI group, F(2, 24)=2.73, p=0.0854. Post-hoc analysis showed significant difference between post-treatment and follow-up, t(24)=−2.34, p=0.0281, d=0.56.
Examination of the full power spectrum more specifically revealed a contiguous range from 15–24Hz with a significant increase at follow-up (Figure 2). Although the magnitude of increase was found to be larger at higher frequencies, these ranges were not statistically significant due to substantial variability. Besides a marginal increase for a narrow subrange at 7–8Hz, no other changes in absolute EEG power were observed. Inspection of individual channels indicated that these increases in beta and gamma power for MBTI were primarily in frontal and central channels (Supplementary Figure 2).
Figure 2.
Combined MBSR and MBTI groups (N=22) demonstrated a significant increase in absolute NREM EEG power primarily overlapping beta frequency band from baseline to 6-month follow-up. Although a larger increase is evident in higher frequencies, these increases are not statistically significant due to substantial variability. As illustrated in the insert, increases in power were present at post-intervention and then amplified at 6-month follow-up. Power values derived from C3/C4 average. Grey error band indicates standard error of the mean (SEM). *p<.05, †p<.10
Exploratory linear mixed models conducted on relative EEG power from baseline to 6-month follow-up (see Supplementary Table 1) demonstrated a significant increase across Time from baseline to follow-up in beta EEG, F(2, 42)=3.38, p=0.0435, and sigma EEG, F(2, 42)=3.71, p=0.0328, which is consistent with the findings for absolute EEG. There was also a decrease in relative delta EEG (1–4Hz), F(2, 42)=3.65, p=0.0346. Post-hoc tests showed this significant decrease occurred in the MBTI group, F(2, 42)=3.23, p=0.0574, but not the MBSR group.
Correlations between EEG and self-report measures
Spearman’s rho correlations were conducted on beta EEG (absolute) and the self-report measures at each time point (see Table 2). At baseline, none of these measures were associated with beta power (p>.10; see Figure 3A). At post-treatment, beta power was negatively correlated with ISI (rho=−0.43, p=0.0474) and PSAS (rho=−0.42, p=0.0507), and positively correlated with FFM (rho=0.37, p=0.0912) (see Figure 4). The relationship between beta power and ISI was also significant at follow-up (rho=−0.43, p=0.0477) (Figure 3B). Beta power was not associated with number of meditation sessions or minutes of meditation practice at either post-treatment or follow-up (p>.10).
Table 2.
Correlations (Spearman’s ρ) between NREM EEG beta power (16–25Hz) and self-report measures at each timepoint.
| Baseline | Post-Treatment | Follow-Up | ||||
|---|---|---|---|---|---|---|
| ρ(22) | p | ρ(22) | p | ρ(22) | p | |
| ISI | −0.01 | 0.9960 | −0.43 | 0.0474 | −0.43 | 0.0477 |
| PSAS | 0.12 | 0.6048 | −0.42 | 0.0507 | −0.07 | 0.7529 |
| GSES | 0.12 | 0.5949 | −0.55 | 0.0080 | −0.16 | 0.4809 |
| HAS | −0.26 | 0.2346 | −0.13 | 0.5538 | −0.34 | 0.1204 |
| FFM | 0.25 | 0.2579 | 0.37 | 0.0912 | 0.33 | 0.1351 |
ISI: Insomnia Severity Index; PSAS: Pre-Sleep Arousal Scale; GSES: Glasgow Sleep Effort Scale; HAS: Hyperarousal Scale; FFM: Five Facet Mindfulness questionnaire.
Figure 3.
Correlation between Insomnia Severity Index (ISI) score and NREM EEG Beta power at Baseline (A) and 6-month follow-up (B) for the combined sample (N=22). Only following mindfulness training did greater Beta power become associated with lower ISI scores (see Table 2), and this relationship was maintained at follow-up.
Figure 4.
Correlation of Five Factor Mindfulness (FFM) scores with NREM EEG Beta power at Post-intervention for the combined sample (N=22). Higher FFM scores were associated with greater EEG Beta power only after mindfulness intervention, similar to the relationship with ISI in Figure 3.
DISCUSSION
The purpose of the present study was to examine EEG sleep microarchitecture of people with insomnia who received a mindfulness-based intervention to generate hypotheses and guide future research regarding objective changes in sleep. Strengths of this study include a well-characterized sample of individuals who met criteria for chronic insomnia disorder and the longitudinal nature of the study, including follow-up data at 6-months post treatment. Our findings revealed several patterns across treatment and follow-up that were most prominent in high frequency EEG. First, we observed an increase in beta EEG (16–25Hz) at both post-treatment and follow-up relative to baseline for those who received an MBI. This increase was driven by MBSR from baseline to post, and then by MBTI from post to follow-up. No significant change from baseline to post-treatment was found in the SM control on either beta or gamma EEG. These findings are consistent with prior reports of meditation and sleep EEG spectral analysis[18,22], providing further support for the possibility that changes in high-frequency EEG activity constitute a potential biological correlate of MBI. However, this contradicts conventional interpretation of beta frequency EEG activity during NREM sleep, which is regarded as a sign of cortical hyperarousal[9,10] that is elevated in insomnia and has been found to be reduced by CBT-I[36]. These paradoxical findings point to a strong need for further research to examine sleep EEG patterns as a function of insomnia treatments.
To provide further context for the EEG patterns, we examined the correlations between beta EEG and self-reported measures, which revealed significant associations only following mindfulness intervention. Specifically, beta EEG power was correlated with the FFM at post-treatment, indicating that higher levels of self-reported mindfulness skills were associated with more beta EEG activity following MBI. Further triangulating the relationship among beta EEG, mindfulness, and insomnia symptoms, ISI scores after intervention (post-treatment and follow-up) was negatively correlated with beta EEG power among those exposed to mindfulness intervention. Together, these results suggest that participation in either MBSR or MBTI can lead to significant improvement in insomnia symptoms, while simultaneously also lead to increases in EEG beta power. Interestingly, beta EEG power was also negatively associated with pre-sleep arousal at post-treatment but not at follow-up. Thus, additional research is warranted to examine if these relationships diminish over time after completion of a mindfulness program, or if long-term effects are specific to global insomnia symptoms but not sleep-related arousal. Further, an important distinction between objective sleep and subjective appraisal is relevant[37], in that reduced insomnia complaints may be due to improvement in nocturnal sleep or daytime functioning, either objectively or subjectively. These results give rise to the hypothesis that reduction in cortical arousal is not necessary for improvement of self-reported insomnia symptoms.
Taken together with our previous findings, the effects of mindfulness meditation on insomnia include a reduction in self-reported arousal and an increase in high-frequency EEG activity during NREM sleep. Although this hypothesis appears paradoxical in the context of current conceptual models of insomnia, potential explanations can be found in the broader literature on mindfulness. Consistent with the metacognitive lens of MBTI, there is an extensive history within the origin traditions of mindfulness and meditation training that emphasizes the cultivation of calm alertness, which in turn enables physiological arousal concurrent with subjective relaxation[17]. From these perspectives, the increase in high-frequency NREM EEG activity is expected, as there is an aim to increase alertness (skill of observation) along with restfulness (experiential peace-of-mind). In this regard, a rather intuitive distinction could be made between ‘adaptive arousal’ and ‘maladaptive arousal’, concepts that have been widely covered in both the clinical and non-clinical literature. Furthermore, it is possible that increases in high-frequency EEG activity during sleep reflect the increased metacognitive activity that allows one to be less distressed about sleep (i.e. reduced insomnia complaints), without changes in traditional PSG-measured sleep parameters. Another interpretation is that participants were able to initiate sleep despite higher cortical arousal or were less likely to awaken in response to cortical activity, resulting in higher beta levels during sleep.
The present study also generated several methodological considerations for conducting spectral EEG analyses on meditation and sleep. First, the specificity of EEG frequency depends on the definitions of the frequency range. The exploratory analyses across EEG bands provide indications that the change in EEG from baseline overlapped the sigma band (primarily 15–16Hz) and the gamma band (25–40Hz) in the MBTI group. While both Ferrarelli et al.[18] and Goldstein et al.[22] found meditation-related changes within the gamma (25–40Hz) range, and there was some indication of increases in this range for MBTI in the present study, the most prominent effect was between 15–24Hz. There appears to be heterogeneity within the insomnia EEG literature, as well, with some reports only finding effects for limited frequency bands above 15Hz, especially when considering sex differences and timecourse across the night[14,38]. Closer inspection of frequency specificity would be useful as this area of research progresses, in order to better delineate the overlap between meditation- and insomnia-related EEG activity. Second, this study primarily focused on absolute (raw) EEG power, while providing relative power (each band divided by total) as a secondary analysis. Although a pattern of increased high-frequency activity similar to absolute power was observed, effects at lower frequency ranges also emerged. Most notably, relative delta/SWA (1–4Hz) was decreased at 6-month follow-up, similar to results reported in Britton et al.[21]. Again, this finding raises a question regarding traditionally understood EEG correlates of insomnia and warrants further investigation. Third, topography is another factor to consider with these findings. Although effects with higher frequency ranges in EEG activity tend to be more diffuse rather than localized, as observed with SWA for example, the previously reported effects with long-term meditators[18] and MBCT[22] were specific to a parietal region of the scalp. Although the present study found similar effects focusing on the nearest available channels at C3/C4, and attenuated effects were also found at F3/F4 (Supplementary Figure 1), it is conceivable that the results would be stronger at P3/P4 locations. Taken together, spectral analysis of EEG appears to be a useful technique for investigating potential neurocognitive effects of mindfulness meditation in sleep but further research is needed to address these methodological issues.
Several limitations of the present study should be considered when interpreting these findings. First, these were secondary analyses from a randomized controlled trial designed to evaluate clinical outcomes rather than quantitative EEG. Moreover, analyses focused on participants with complete data to mitigate missing data issues with statistical analysis. Therefore, it is possible that these analyses are biased towards completers of MBSR and MBTI and might not represent all of those who attend a mindfulness program. Furthermore, power considerations were based on the main clinical outcomes and not the EEG variables used in this study. In addition, PSG data were obtained for only one night at each time point and could reflect sleeping in a different environment, although “first night effects” are unlikely to be an issue given that participants underwent a screening PSG prior to baseline and EEG spectral analysis tends to be stable across nights[39]. Also, meditation practice was measured using self-report, which is subject to bias and over-reporting. This should be noted when interpreting the relationship between EEG and mindfulness measures, since the correlation between EEG beta activity and FFM was significant only at post-intervention, but no significant correlation was found with mindfulness practice. Finally, since MBTI also includes some behavioral strategies for insomnia, it cannot be determined if the EEG changes in this group are due to meditation practice or use of the behavioral strategies.
Conclusions
The findings of this hypothesis-generating study give rise to the hypothesis that changes in high-frequency sleep EEG power is a potential biological marker associated with mindfulness training. Also, the hypothesis that reduction in cortical arousal (i.e., high-frequency EEG) is not necessary for improvement of self-reported insomnia symptoms merits further testing. Future studies should test these hypotheses by using rigorous designs guided by the methodological considerations discussed for conducting EEG spectral analyses in meditation and sleep.
Supplementary Material
Supplementary Figure 1. Individual spectra (absolute power, C3/C4 average across NREM sleep) for MBSR (A) and MBTI (B) separately with time-point averages overlaid.
Supplementary Figure 2. Changes (% from Baseline) in NREM EEG Beta range (16–25Hz) power at 6-month follow-up for individual channels and mindfulness groups plotted separately. *p<.05 (uncorrected)
Highlights:
NREM EEG beta (16–25Hz) power during sleep increased following mindfulness-based interventions.
Pre-treatment, this NREM EEG activity was not associated with mindfulness or insomnia.
Post-treatment, correlations emerged, in which NREM EEG beta power was positively associated with mindfulness and negatively associated with insomnia.
Conflicts of Interest and Source Funding
This research project was supported by a grant from the National Institutes of Health, National Center for Complementary and Integrative Health (K23AT003678 to JCO), and by a National Science Foundation Graduate Research Fellowship (to MRG). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
REFERENCES
- [1].Davidson RJ, Kabat-Zinn J, Schumacher J, Rosenkranz M, Muller D, Santorelli SF, Urbanowski F, Harrington A, Bonus K, Sheridan JF, Alterations in brain and immune function produced by mindfulness meditation., Psychosom. Med 65 (2003) 564–70. http://www.ncbi.nlm.nih.gov/pubmed/12883106 (accessed July 20, 2018). [DOI] [PubMed] [Google Scholar]
- [2].Black DS, Slavich GM, Mindfulness meditation and the immune system: a systematic review of randomized controlled trials, Ann. N. Y. Acad. Sci 1373 (2016) 13–24. doi: 10.1111/nyas.12998. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [3].Carlson L, Rouleau CR, Garland SN, The impact of mindfulness-based interventions on symptom burden, positive psychological outcomes, and biomarkers in cancer patients, Cancer Manag. Res 7 (2015) 121. doi: 10.2147/CMAR.S64165. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [4].Buysse DJ, Insomnia., JAMA 309 (2013) 706–16. doi: 10.1001/jama.2013.193. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [5].Qaseem A, Kansagara D, Forciea MA, Cooke M, Denberg TD, Management of Chronic Insomnia Disorder in Adults: A Clinical Practice Guideline From the American College of Physicians, Ann. Intern. Med 165 (2016) 125. doi: 10.7326/M15-2175. [DOI] [PubMed] [Google Scholar]
- [6].Winbush NY, Gross CR, Kreitzer MJ, The effects of mindfulness-based stress reduction on sleep disturbance: a systematic review., Explore (NY) 3 (2007) 585–91. doi: 10.1016/j.explore.2007.08.003. [DOI] [PubMed] [Google Scholar]
- [7].Gong H, Ni C-X, Liu Y-Z, Zhang Y, Su W-J, Lian Y-J, Peng W, Jiang C-L, Mindfulness meditation for insomnia: A meta-analysis of randomized controlled trials, J. Psychosom. Res 89 (2016) 1–6. doi: 10.1016/j.jpsychores.2016.07.016. [DOI] [PubMed] [Google Scholar]
- [8].Ong JC, Smith CE, Using Mindfulness for the Treatment of Insomnia, Curr. Sleep Med. Reports 3 (2017) 57–65. doi: 10.1007/s40675-017-0068-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [9].Riemann D, Spiegelhalder K, Feige B, Voderholzer U, Berger M, Perlis M, Nissen C, The hyperarousal model of insomnia: a review of the concept and its evidence., Sleep Med. Rev 14 (2010) 19–31. doi: 10.1016/j.smrv.2009.04.002. [DOI] [PubMed] [Google Scholar]
- [10].Bonnet MH, Arand DL, Hyperarousal and insomnia: State of the science, Sleep Med. Rev 14 (2010) 9–15. doi: 10.1016/j.smrv.2009.05.002. [DOI] [PubMed] [Google Scholar]
- [11].Perlis ML, Merica H, Smith MT, Giles DE, Beta EEG activity and insomnia., Sleep Med. Rev 5 (2001) 363–374. http://www.ncbi.nlm.nih.gov/pubmed/12531000 (accessed November 25, 2013). [DOI] [PubMed] [Google Scholar]
- [12].Perlis ML, Smith MT, Andrews PJ, Orff H, Giles DE, Beta/Gamma EEG activity in patients with primary and secondary insomnia and good sleeper controls., Sleep 24 (2001) 110–7. http://www.ncbi.nlm.nih.gov/pubmed/11204046 (accessed May 22, 2013). [DOI] [PubMed] [Google Scholar]
- [13].Perlis ML, Kehr EL, Smith MT, Andrews PJ, Orff H, Giles DE, Temporal and stagewise distribution of high frequency EEG activity in patients with primary and secondary insomnia and in good sleeper controls., J. Sleep Res 10 (2001) 93–104. http://www.ncbi.nlm.nih.gov/pubmed/11422723 (accessed April 29, 2013). [DOI] [PubMed] [Google Scholar]
- [14].Spiegelhalder K, Regen W, Feige B, Holz J, Piosczyk H, Baglioni C, Riemann D, Nissen C, Increased EEG sigma and beta power during NREM sleep in primary insomnia., Biol. Psychol 91 (2012) 329–33. doi: 10.1016/j.biopsycho.2012.08.009. [DOI] [PubMed] [Google Scholar]
- [15].Riedner BA, Goldstein MR, Plante DT, Rumble ME, Ferrarelli F, Tononi G, Benca RM, Regional patterns of elevated alpha and high-frequency electroencephalographic activity during nonrapid eye movement sleep in chronic insomnia: A pilot study, Sleep 39 (2016) 801–812. doi: 10.5665/sleep.5632. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [16].Krystal AD, Edinger JD, Sleep EEG predictors and correlates of the response to cognitive behavioral therapy for insomnia., Sleep 33 (2010) 669–77. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2864882&tool=pmcentrez&rendertype=abstract (accessed August 5, 2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
- [17].Britton WB, Lindahl JR, Cahn BR, Davis JH, Goldman RE, Awakening is not a metaphor: the effects of Buddhist meditation practices on basic wakefulness., Ann. N. Y. Acad. Sci 1307 (2014) 64–81. doi: 10.1111/nyas.12279. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [18].Ferrarelli F, Smith R, Dentico D, Riedner BA, Zennig C, Benca RM, Lutz A, Davidson RJ, Tononi G, Experienced mindfulness meditators exhibit higher parietal-occipital EEG gamma activity during NREM sleep., PLoS One 8 (2013) e73417. doi: 10.1371/journal.pone.0073417. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [19].Lutz A, Greischar LL, Rawlings NB, Ricard M, Davidson RJ, Long-term meditators self-induce high-amplitude gamma synchrony during mental practice, Proc. Natl. Acad. Sci 101 (2004) 16369–16373. doi: 10.1073/pnas.0407401101. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [20].Cahn BR, Polich J, Meditation states and traits: EEG, ERP, and neuroimaging studies., Psychol. Bull 132 (2006) 180–211. doi: 10.1037/0033-2909.132.2.180. [DOI] [PubMed] [Google Scholar]
- [21].Britton WB, Haynes PL, Fridel KW, Bootzin RR, Polysomnographic and subjective profiles of sleep continuity before and after mindfulness-based cognitive therapy in partially remitted depression., Psychosom. Med 72 (2010) 539–48. doi: 10.1097/PSY.0b013e3181dc1bad. [DOI] [PubMed] [Google Scholar]
- [22].Goldstein MR, Britton WB, Allen JJB, Bootzin RR, Effects of a mindfulness-based depression relapse prevention program on quantitative sleep EEG, in: SLEEP, 29th Annu. Meet. Assoc. Prof. Sleep Soc., Seattle, 2015. [Google Scholar]
- [23].Ong J, Manber R, Segal Z, Xia Y, Shapiro S, Wyatt J, A Randomized Controlled Trial of Mindfulness Meditation for Chronic Insomnia, Sleep 37 (2014) 1553–1563. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [24].Ong JC, Xia Y, Smith-Mason CE, Manber R, A Randomized Controlled Trial of Mindfulness Meditation for Chronic Insomnia: Effects on Daytime Symptoms and Cognitive-Emotional Arousal, Mindfulness (N. Y) (2018) 1–11. doi: 10.1007/s12671-018-0911-6. [DOI] [Google Scholar]
- [25].American Academy of Sleep Medicine, The International Classification of Sleep Disorders, Second Edition (ICSD-2), 2005.
- [26].Kabat-Zinn J, Full catastrophe living : using the wisdom of your body and mind to face stress, pain, and illness, n.d. [Google Scholar]
- [27].Carskadon M, Rechtschaffen A, Monitoring and staging human sleep, in Principles and practice of sleep medicine, in: Kryger M, Roth T, Dement W (Eds.), Princ. Pract. Sleep Med., Elsevier, Philadelphia, 2005: pp. 1359–1377. [Google Scholar]
- [28].Iber C, Ancoli-Israel S, Chesson A, Quan S, American Academy of Sleep Medicine, he AASM Manual for the Scoring of Sleep and Associated Events:Rules, Terminology, and Technical Specifications First edition., Westchester, IL, 2007. [Google Scholar]
- [29].Goldstein MR, Cook JD, Plante DT, The 5α-reductase inhibitor finasteride is not associated with alterations in sleep spindles in men referred for polysomnography, Hum. Psychopharmacol 31 (2016) 70–74. doi: 10.1002/hup.2502. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [30].Plante DT, Goldstein MR, Medroxyprogesterone acetate is associated with increased sleep spindles during non-rapid eye movement sleep in women referred for polysomnography., Psychoneuroendocrinology 38 (2013) 3160–3166. doi: 10.1016/j.psyneuen.2013.08.012. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [31].Bastien CH, Vallières A, Morin CM, Validation of the Insomnia Severity Index as an outcome measure for insomnia research., Sleep Med 2 (2001) 297–307. http://www.ncbi.nlm.nih.gov/pubmed/11438246 (accessed July 20, 2018). [DOI] [PubMed] [Google Scholar]
- [32].Nicassio PM, Mendlowitz DR, Fussell JJ, Petras L, The phenomenology of the pre-sleep state: the development of the pre-sleep arousal scale., Behav. Res. Ther 23 (1985) 263–71. http://www.ncbi.nlm.nih.gov/pubmed/4004706 (accessed July 20, 2018). [DOI] [PubMed] [Google Scholar]
- [33].Broomfield NM, Espie CA, Towards a valid, reliable measure of sleep effort., J. Sleep Res 14 (2005) 401–7. doi: 10.1111/j.1365-2869.2005.00481.x. [DOI] [PubMed] [Google Scholar]
- [34].Pavlova M, Berg O, Gleason R, Walker F, Roberts S, Regestein Q, Self-reported hyperarousal traits among insomnia patients, J. Psychosom. Res 51 (2001) 435–441. doi: 10.1016/S0022-3999(01)00189-1. [DOI] [PubMed] [Google Scholar]
- [35].Baer RA, Smith GT, Hopkins J, Krietemeyer J, Toney L, Using Self-Report Assessment Methods to Explore Facets of Mindfulness, Assessment 13 (2006) 27–45. doi: 10.1177/1073191105283504. [DOI] [PubMed] [Google Scholar]
- [36].Cervena K, Dauvilliers Y, Espa F, Touchon J, Matousek M, Billiard M, Besset A, Effect of cognitive behavioural therapy for insomnia on sleep architecture and sleep EEG power spectra in psychophysiological insomnia., J. Sleep Res 13 (2004) 385–93. doi: 10.1111/j.1365-2869.2004.00431.x. [DOI] [PubMed] [Google Scholar]
- [37].Lichstein KL, Insomnia identity., Behav. Res. Ther 97 (2017) 230–241. doi: 10.1016/j.brat.2017.08.005. [DOI] [PubMed] [Google Scholar]
- [38].Buysse DJ, Germain A, Hall ML, Moul DE, Nofzinger EA, Begley A, Ehlers CL, Thompson W, Kupfer DJ, EEG spectral analysis in primary insomnia: NREM period effects and sex differences., Sleep 31 (2008) 1673–82. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2603490&tool=pmcentrez&rendertype=abstract (accessed May 22, 2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
- [39].Israel B, Buysse DJ, Krafty RT, Begley A, Miewald J, Hall M, Short-term stability of sleep and heart rate variability in good sleepers and patients with insomnia: for some measures, one night is enough., Sleep 35 (2012) 1285–91. doi: 10.5665/sleep.2088. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supplementary Figure 1. Individual spectra (absolute power, C3/C4 average across NREM sleep) for MBSR (A) and MBTI (B) separately with time-point averages overlaid.
Supplementary Figure 2. Changes (% from Baseline) in NREM EEG Beta range (16–25Hz) power at 6-month follow-up for individual channels and mindfulness groups plotted separately. *p<.05 (uncorrected)