The Difference in Spectral Power Density in Sleep Electroencephalography According to the Estimation of Total Sleep Time
Article information
Abstract
Objective
The discrepancy between self-reported subjective sleep time and objective sleep time varies from person to person, and the mechanism that can explain this discrepancy is unclear. This study aimed to investigate the difference in the spectral power of sleep electroencephalography (EEG) according to the estimation of total sleep time (TST).
Methods
Of the 4,080 participants in the Sleep Heart Health Study, 2,363 participants with complete data from polysomnography, morning sleep survey, and spectral power of sleep EEG were included in the analysis. The participants were classified as normo-estimators (estimation of TST <±60 min), under-estimators (underestimation of TST ≤60 min), or over-estimators (overestimation of TST ≥60 min). A fast Fourier transformation was used to calculate the EEG power spectrum for total sleep duration within contiguous 30-s epochs of sleep. The sleep EEG spectral power was compared among the groups after adjusting for potential confounding factors, such as age, sex, proportion of participants with insomnia, apnea-hypopnea index, and TST.
Results
Of the 2,363 participants, 1,507 (63.8%), 412 (17.4%), and 444 (18.8%) were assigned to the normo-, under-, and over-estimator groups, respectively. The power spectra during total sleep differed significantly among the groups in the delta (p=0.008) and theta bands (p=0.017) after controlling for potential confounders.
Conclusion
Higher delta and theta powers were found in the under-estimators than in the over-estimators. This study suggests that differences exist in the microstructures of polysomnography-derived sleep EEG between these two groups. This study suggests that differences exist in the microstructures of polysomnography-derived sleep EEG based on differences in sleep time estimation.
INTRODUCTION
The estimation of quantity and quality of sleep is subjective. Therefore, subjective sleep duration may differ from objective measurements obtained using polysomnography (PSG) [1]. Some individuals who sleep for >6 h may report not sleeping at all [2], whereas patients with insomnia often underestimate their total sleep time (TST) [3-6]. People with sleep state misperception, also known as paradoxical insomnia, complain of insomnia without an evidence of a short sleep duration [7-10]. Underestimation of sleep is also common in the general population, especially in middle-aged adults [8,11]. Previous researchers have found that individuals with psychiatric disorders including depression and diverse medical conditions such as rheumatoid arthritis, frequently underestimate their sleep [12-17].
Conversely, others might report that they slept far more than their objective sleep time [11,18,19]. Such sleep overestimation seems to be prevalent; however, this has been relatively less reported. Previous demographic studies have consistently shown that people in the general population, especially those with short sleep durations [20], tend to overestimate their sleep [11,21,22].
The etiology of differences in sleep estimations is unclear; however, previous studies have suggested abnormalities in sleep electroencephalography (EEG), neuroendocrine function, metabolism, and activity in the important sleep-wake regulating brain regions such as the ventrolateral preoptic nucleus and limbic system [23-26]. Studies of sleep EEG have shown that differences or alterations in EEG during sleep and the microstructure of sleep can influence sleep estimation [27-29]. However, few have addressed the difference in sleep EEG according to the estimation of sleep time. Previously, we studied the relationship between subjective-objective discrepancy (SOD) and quality of life in a large data set, the Sleep Heart Health Study (SHHS) [30]. The SHHS is a qualified large cohort study equipped with data on clinical characteristics, subjective sleep time, PSG (objective sleep data), and spectral power of sleep EEG. Therefore, this data is suitable for studying the differences in spectral power of sleep EEG according to differences in sleep estimation.
This study aimed to find the difference in spectral power of sleep EEG according to the estimation of TST in a communitydwelling population.
METHODS
Study cohort and characteristics of the data
The SHHS is an open, prospective cohort study with a largescale, multi-center, and community-based dataset. It intended to study cardiovascular diseases and other conditions, such as hypertension, cerebrovascular disease, coronary heart disease, and mortality of sleep apnea among the population in the United States [31,32]. After SHHS-1, a second study that included PSG data from 4,586 SHHS participants (SHHS-2) was conducted. All participants provided written informed consent, and the study was approved by the Institutional Review Board (IRB) of each participating institution (ClinicalTrials.gov Identifier: NCT00005275).
We used a subset of the SHHS 2 database obtained from the National Sleep Research Resource (https://sleepdata.org/datasets/shhs). We accessed the SHHS database after acquiring a signed agreement with Brigham and Women’s Hospital. Our project was exempted from review by the IRB of Gil Medical Center (IRB No. GDIRB2018-005). This data included the clinical characteristics of the participants and sleep habit information, including frequency of insomnia, PSG measurements, and subjective sleep reports, based on a morning survey about the previous night’s sleep after PSG. The participants with insomnia symptoms (i.e., difficulty falling asleep, difficulty staying asleep, or waking up too early) occurring on ≥16 nights per month were considered as having insomnia. Participants with complete PSG data, subjective TST data measured in the morning sleep survey, and PSG with spectral power of sleep EEG were finally included in the study.
PSG
All participants underwent overnight PSG at home, as previously described, using a Compumedics P-series recording system (Compumedics, Abbotsford, Australia) [33]. The recording montage consists of a C3–A2 and C4–A1 electroencephalogram, chin electromyogram, two electrooculograms, a single-lead electrocardiogram, an airflow sensor (an oral-nasal thermistor), oxyhemoglobin saturation measurement by pulse oximetry, measurement of respiratory effort using impedance plethysmography, and data on body position. Sleep stage scoring of all nocturnal recordings was conducted by trained technicians at a centralized PSG reading center using the Rechtschaffen and Kales criteria [34]. Apnea was defined as absent or nearly absent airflow lasting for more than 10 s [35]. Hypopnea was identified as a discernible reduction lasting more than 10 s in airflow or thoracoabdominal movement (≥30% below the baseline values) [35]. As these data had various apnea–hypopnea index (AHI) based on different hypopnea criteria, we adopted the AHI scored based on the same criteria as recommended by the hypopnea rule of the 2007 version of the American Academy of Sleep Medicine manual [9].
Spectral analysis
The spectral analysis method used has been described in a previous study [36]. The C3–A2 and C4–A1 EEG recordings were sampled at 125 Hz and analyzed using the fast Fourier transform. The fast Fourier transform was conducted on a 5-s EEG segment to obtain a frequency resolution of 0.2 Hz. Each 5-s EEG segment was first windowed with a Hanning tapering window before computing the power spectra. The power content expressed as μV2 for each 30-s epoch of sleep was determined as the average power across the six 5-s segments of the sleep EEG. Power spectra were computed for each EEG frequency band: slow oscillation (0.5–1 Hz), delta (1–4 Hz), theta (4–8 Hz), alpha (8–12 Hz), sigma (12–15 Hz), and beta (15–20 Hz). Frequencies <0.8 Hz were excluded from the analyses to control for low-frequency artifacts, such as those caused by sweating and respiration [36]. Data derived from the central EEG electrodes (i.e., [C3/A2 + C4/A1]/2) recorded over the total sleep period were used for the present analysis. For the analysis, we used absolute spectral power, which is the integral of all the power values within each frequency range. To achieve normal distributions, all absolute power data were log-transformed [37].
Group classification after the calculation of the SOD
To calculate the SOD of total sleep duration, the objective TST measured using PSG was subtracted from the subjective TST, which was recorded in the morning sleep survey after the PSG. A difference of ≥60 min in the TST was defined as marked SOD in the previous literature [7,20]. Based on this criteria, participants were classified into the following three groups: 1) the normo-estimator group (SOD, <±60 min), 2) the under-estimator group (SOD ≤-60 min), and 3) the over-estimator group (SOD ≥60 min).
Statistical analysis
Demographic, clinical, and PSG data were compared among the three groups using an analysis of variance. To compare the spectral power of sleep EEG among the groups, we employed analysis of covariance, controlling for age, sex, proportion of participants with insomnia, AHI, and TST measured using PSG, and post-hoc testing for Bonferroni-adjusted multiple comparisons. Statistical analysis was performed at a two-sided significance level of p<0.05 using SPSS for Windows (version 23, IBM Corp., Armonk, NY, USA).
RESULTS
In total, 2,363 participants were included in the final study population. Table 1 shows the demographic and clinical characteristics and PSG results. The average age of the participants was 62.2 years; 1,057 (44.7%) were men. The mean subjective TST measured using the morning sleep survey was 379.1 min, and the mean SOD of TST of all participants was 3.2 min. In PSG, the mean sleep efficiency was 79.5%, and the mean AHI was 16.0 per hour.
Demographic and polysomnographic data of participants
Table 2 shows the clinical characteristics and PSG measurements of three groups according to sleep estimation and their inter-group differences. One thousand five hundred seven participants were normo-estimators (63.8%), 412 were under-estimators (17.4%), and 444 were over-estimators (18.8%). In the over-estimator group, the age and proportion of men were higher (p<0.001) than those in the other groups. In addition, the under-estimator group had the highest proportion of participants with insomnia (p<0.001).
Comparison of demographic and PSG data among the three groups of participants
The over-estimator group had the lowest sleep efficiency (p< 0.001), longest sleep latency (p<0.001), highest wake after sleep onset (p<0.001), and highest AHI (p<0.001) among all the groups. They also had the highest proportion of stage N1 (p<0.001) and N2 (p<0.001) and the lowest proportion of stage N3 (p<0.001) and R (p<0.001) among the groups, in terms of sleep architecture measured using PSG. In contrast, the under-estimator group had the longest TST measured by PSG (p<0.001) and longest time in bed (p<0.001) among all the groups.
The log-transformed spectral power of sleep EEG were compared among the three groups after controlling for potential confounding factors (age, sex, frequency of participants with insomnia, AHI, and TST), and this was significantly different among the groups (Table 3). The delta power of the under-estimator group was significantly higher than that of the over-estimator group (p=0.008). Additionally, the theta power of the under-estimator group was significantly higher than that of the over-estimator group (p=0.017).
DISCUSSION
This study found that the under-estimator group had higher delta and theta power (considered slow waves) than the over-estimator group did. In this study, subjective TST was significantly different among the over-estimators (447.4 min), normo-estimators (384.4 min), and under-estimators (286.0 min). In comparison, the actual TST measured by PSG was significantly different among the over-estimators (337.9 min), normo-estimators (380.7 min), and under-estimators (399.7 min), with the under-estimators having the longest sleep duration. The under-estimators had a higher sleep efficiency and a higher proportion of N3 (deep sleep) than normo-estimators and over-estimators did, not only in terms of TST but also the sleep structures measured by PSG. This seems to be consistent with the main results of this study (i.e., greater spectral power of slow-wave EEG in under-estimators).
To the best of our knowledge, only one previous study, by Lecci et al. [38], on the spectral power of sleep EEG according to the sleep estimation in the general population has been reported. They showed that the delta power was decreased and the beta power was increased in under-estimators, which is contrary to our results [38]. This inconsistent result might be due to the differences in study methods. Their study adopted the concept of Sleep Perception Index to arbitrarily categorize the top and bottom 2.5% of the participants as over-estimators and under-estimators, respectively [38]. In our data from the SHHS-2, the mean subjective TST was 379.1 min, TST measured by PSG was 375.9 min, and the mean SOD of TST was 3.2 min, indicating only a minor discrepancy. This suggests that the data in our study have a low chance of systematic error in sleep estimation.
Although our study controlled for age, sex, AHI, and TST among the three groups, other clinical characteristics between the three groups might have acted as confounding factors. The under-estimator group was younger than the over-estimator group and had fewer male participants. Deep sleep decreases as humans age; however, a previous study showed that slow-wave sleep tends to decrease more significantly with age in men than in women [39]. In addition, there was a significant difference in the proportion of participants with insomnia, AHI, and TST between the under-estimators and over-estimators. These differences in clinical characteristics between the groups might have created a bias, although these potential confounding factors were controlled for in the statistical analysis.
The findings of this study showed a greater spectral power of the slow-wave EEG band in the under-estimator group than in the over-estimator group. As our results contradict those reported in previous studies, more replication studies on this topic and establishment of data with matched clinical characteristics among various groups are needed.
Notes
Funding Statement
The Sleep Heart Health Study (SHHS) was supported by National Heart, Lung, and Blood Institute cooperative agreements U01HL53916 (University of California, Davis), U01HL53931 (New York University), U01HL53934 (University of Minnesota), U01HL53937 and U01HL64360 (Johns Hopkins University), U01HL53938 (University of Arizona), U01HL53940 (University of Washington), U01HL53941 (Boston University), and U01HL63463 (Case Western Reserve University). The National Sleep Research Resource was supported by the National Heart, Lung, and Blood Institute (R24 HL114473, RFP 75N92019R002). This work was also supported by the National Research Foundation of Korea (NRF) grant funded by the Korean government (MSIT) (grant number: NRF-2020R1A2C1007527).
The authors have no potential conflicts of interest to disclose.
Availability of Data and Material
The datasets generated or analyzed during the current study are available in the National Sleep Research Resource (https://sleepdata.org/datasets/shhs) after acquiring a signed agreement with Brigham and Women’s Hospital.
Author Contributions
Conceptualization: Seung-Gul Kang. Data curation: Seung-Gul Kang, Jun-Mu Heo. Formal analysis: all authors. Methodology: Seung-Gul Kang, Seo-Eun Cho. Software: Seung-Gul Kang. Validation: Seung-Gul Kang. Writing—original draft: all authors. Writing—review & editing: Seung-Gul Kang.