Effect of napping on a bean bag chair on sleep stage, muscle activity, and heart rate variability

Ethics statement

The research protocols were aapproved by the Academic Research Ethical Review Committee of Waseda University (IRB #2019-193) and the work was performed in accordance with the 1964 Declaration of Helsinki. Written informed consent was obtained from all participants before participating in experiments.

Materials

Figure 1 shows the bedding and lying posture used in this study. For the BC condition, a BC filled with soft micro-beads was used (Yogibo Max^®; Webshark Inc., Osaka, Japan). For the UR condition, a UR-based, polyethylene-surfaced cushion chair (LOWYA^® Long seat chair; Vega Co., Fukuoka, Japan) was used. The chair sizes were quite similar (BC: 170 cm × 48 cm × 70 cm, UR:172 cm × 56 cm × 75 cm), and the BC was 16% lighter than the UR (BC: 8.0 kg, UR: 11.5 kg). The chairs were installed in a quiet and dark shield room in the sleep laboratory of Waseda University, where the temperature was set at approximately 25.0 °C and the light intensity level was adjusted to approximately 200 lx, as measured with a light meter.

Participants and setting

Fourteen healthy participants (two females and 12 males, age: 22.5 ± 1.5 years) with normal sleep habits were enrolled. The sample size calculation was determined using G*Power 3.1.9.6 statistical power analysis software (Faul et al., 2007). Power analysis indicated that a sample size of 12 per group (number of measurement = 2) were needed for a η_p² (0.25) when α = 0.05 for a power of 0.60 with two independent conditions, using repeated measures of variance (ANOVA), within -between interactions. Statistical significance was set at p < 0.05 (two-tailed).

Participants were recruited through poster advertisements, social media, and emails to students of Waseda University. Participants were selected based on the absence of any prior diagnosis of sleep disorders and them not having taken any medication and alcohol at the time of the experiments. Participants were asked to maintain regular sleep schedule for 1 week before the laboratory experiment.

All participants were assessed using the Pittsburgh Sleep Quality Index (PSQI) for habitual sleep quality and Morningness-Eveningness Questionnaire (MEQ) for individual circadian preference. MEQ was a self-assessment questionnaire to determine morningness and eveningness, representing the degree to which respondents are active and alert at certain times of day (Horne & Ostberg, 1976). Subjective bedding preferences were also assessed using the visual analogue scale (VAS; higher scores = softer bedding preference). The mean values for the participants’ baseline assessment were as follows: body mass index, 22.7 ± 1.8; subjective bedding preferences, 4.9 ± 1.8; PSQI, 5.8 ± 2.8; and MEQ, 43.3 ± 9.6.

Study design

Participants were enrolled within the context of a randomized, single-blind, crossover study to evaluate the effects of BC and UR at a minimum of 1-week intervals (Fig. 2). Participants entered the laboratory and were asked to rate their level of sleepiness based on the KSS before and after the nap. Participants reported subjective evaluation of sleep in VAS scores after taking a nap.

Participants were not informed about which type of chair they would be napping in. Participants were requested to sit on the experimental chair for napping after the lights were switched off at 13:00. Participants were woken up as lights were turned back on at 14:00 by the examiner.

Actigraphy recording

To monitor sleep in the night preceding the experiment, participants were requested to wear an MTN-220 (Acos Co., Ltd., Nagano, Japan) on the front side of the trunk by clipping it to their waist belt or the edge of their trousers/pants. This actigraph is a small and light (9 g) coin-shaped device (external dimensions of 27 mm in diameter and 9.8 mm in depth, including the clip) that records the amount of physical activity by employing an internal three-axis accelerometer, enabling sleep evaluation at home prior to the experiment. Recording started at 18:00 the preceding night and stopped at 08:00 on the day of the experiment. The actigraph was removed at 8:00 when the recording ended and was not used for the napping experiment. The sensitivity and specificity of MTN-220 were equivalent to those determined for conventional actigraphy (Nakazaki et al., 2014) and PSG (Enomoto et al., 2009). Nocturnal sleep propensity on the night before the experiment was compared between the two bedding conditions.

Electroencephalogram and sleep stage scoring

EEG recordings were obtained during the nap using a portable recording system (Brainwave Sensor, ZA-X^®; Proassist, Ltd., Osaka, Japan). This system consisted of a wireless transmitter and receiver, using a pair of bipolar EEG and EOG electrode leads connected to the transmitter. The transmitter is 65 mm (W) × 36 mm (L) × 16 mm (H) and weighs 22 g, while the receiver is 135 mm (W) × 76 mm (L) × 27 mm (H) and weighs 135 g. Disposable electrodes (Vitrode F-150M; Nihon Koden Corp., Tokyo, Japan) for channel 1 were placed 1 cm above and lateral to the outer canthus of the left eye and right mastoid. Two other disposable electrodes connected to channel 2 were placed 1 cm below and slightly lateral to the outer canthus of the right eye as well as above the chin.

The signals were recorded at a sampling rate of 128 Hz with filters of 0.5–40 Hz for EEG and 0.5–10 Hz for EOG. Sleep stages were scored every 20 s by a clinical professional technologist without knowledge of the interventions according to the American Academy of Sleep Medicine (AASM) manual scoring rules (Berry et al., 2020). Although the 20 s epoch was shorter than that officially defined in AASM scoring rule, applying a shorter epoch provides more accurate validity in determining sleep stages (Schulz, 2008). Sleep stages were classified into the four sleep stages: (1) awake, (2) rapid eye movement (REM) sleep, (3) light non-REM (NREM) sleep (stages N1 and N2), and (4) deep NREM sleep (N3). To validate, simultaneous sleep recordings using the portable two-channel device showed robust agreement in sleep scoring with PSG; for example, kappa values were 0.80 overall and inter-scorer concordance rates were 60.1% in the two-channel and 71.7% in the PSG (Nonoue et al., 2017).

EEG recordings during sleep were classified into 20-s intervals and brain waves were categorized as follows: delta (1.0–4.0 Hz), theta (4.0–8.0 Hz), alpha (8.0–12.0 Hz), and beta (16.0–35.0 Hz). Fast Fourier Transform was performed to provide the mean spectral density on extracted intervals to assess the component of each frequency band.

Electromyography

Continuous EMG recording of body muscles during the nap was carried out, minimizing the number of muscles attached to the channel to avoid disturbing sleep. The muscle activity of the bilateral trapezius muscle and the bilateral abdominal oblique external muscle were measured. Two EMG probes were positioned 2 cm apart on the bilateral external oblique muscle with Ag/AgCl electrodes. Both electrodes were placed around the center of a line drawn between the inferior borders of the 10th rib and pubis and on the muscle belly under the 10th rib. For the trapezius muscle, probes were placed 2 cm apart on the upper fibers of the trapezius, approximately 2 cm to the lateral inferior direction of the vertebra prominens (seventh cervical vertebra, C7). Additionally, referenced probes were also placed on the left shoulder, superiorly.

For the measurement of EMG signals, a multi-telemeter system (Polymate mini A108; Miyuki Giken, Co. Ltd., Tokyo) was employed. Raw EMG signals were sampled at 500 Hz, with a time constant of 0.01 s, and notch-filtered to erase the 50 Hz frequency interference. For the EMG amplification, the high cutoff filter was set to 300 Hz and the low cutoff filter to 30 Hz to exclude frequency relevant to respiratory and any heart- or motion-associated activity (Abbaspour & Fallah, 2014; Chowdhury et al., 2013).

To assess the activity of superficial muscles in the EMG signal in the time domain, an integrated EMG (iEMG) was calculated. The iEMG is the mathematical integration of a rectified electrical muscle stimulation signal, representing a method for numerical integration and numerical approximation of definite integrals (Truong Quang Dang et al., 2012). Specifically, it is the following approximation:

$$\mathrm{Y}=\underset{\mathrm{a}}{\overset{\mathrm{b}}{\int }}\mathrm{f}\left(\mathrm{x}\right)\mathrm{d}\mathrm{x}\approx {\displaystyle \frac{\mathrm{b}-\mathrm{a}}{6}\left[\mathrm{f}\left(\mathrm{a}\right)+4\mathrm{f}\left({\displaystyle \frac{\mathrm{a}+\mathrm{b}}{2}}\right)+\mathrm{f}\left(\mathrm{b}\right)\right]}$$

In this formula, a and b are the unit spacing. F(a) and f(b) are the amplitudes of the EMG signal at a and b. The iEMG was computed as an approximation of the cumulative integral of the whole EMG signal amplitude during a 20 s period.

Heart rate variability

Actiheart 5 monitors (Cambridge Neurotechnology, Cambridge, UK) were used to record heart interbeat interval (IBI) data to evaluate HRV. These are flat and lightweight (10.5 g) wearable devices for which high levels of intra- and inter-instrument reliability, as well as good validity of measures, have been reported (Brage et al., 2005).

Participants were requested to attach the Actiheart monitor on their chest with a fastened belt during their nap. A 15-min period of properly collected data during the sleep period was extracted for analysis by visually screening the raw data. More information on how the ECG IBI artifacts were discarded/removed is necessary. Artifacts in IBI data detected visually were removed and the resulting data were concatenated. Expressed IBI data indicated as invalid data according to the software (Actiheart Software ver.5; Cambridge Neurotechnology, Cambridge, UK) were also excluded from the analysis.

In terms of frequency analysis, the integrated spectral power of the periodic components was calculated to quantify the total spectral power (TP; ≤0.4 Hz; ms²) and the power in the low-frequency (LF; 0.04–0.15 Hz; ms²) and high-frequency (HF; 0.15–0.4 Hz; ms²) bands. The HF normalized units (HFnu = HF/(HF + LF) × 100) and the LF/HF ratio were also computed. Furthermore, for the time domain analysis, the root mean square of successive differences (RMSSD) between adjacent R-R intervals was also calculated. We applied HFnu and RMSSD as indices for parasympathetic tone and LF/HF, which represents the balance between the sympathetic and parasympathetic systems, reflecting sympathetic dominance when it is a higher value (Djaoui et al., 2017; Laborde, Mosley & Thayer, 2017).

For the analysis of IBI data and frequency-domain HRV measures during different sleep stages, 3-min artifact-free windows were selected during the nap. The IBI data obtained during sleep were extracted based on the sleep stages classified as mentioned above. Sleep stages data to be extracted were N1 and N2 only; N3 stage sleep was excluded because only some participants showed signs of N3 stage sleep. Additionally, a 3-min pre-nap window was analyzed as the wakefulness period. The IBI data obtained during N1 and wake after sleep onset were excluded from the analysis due to difficulty in isolating the recording as a result of instability.

Fatigue assessment

To assess the effects of napping on recovery from fatigue, a flicker test was conducted after napping. The flicker test was established to measure the flicker perception threshold, which changes as a result of the level of arousal due to fatigue. We evaluated critical flicker–fusion frequency (CFF), reflective of fatigue and mental workload (Luczak & Sobolewski, 2005). To avoid sleep inertia, the flicker test was conducted at least 30 min after awakening from the napping experiment.

Statistical analysis

For demographic variables, we calculated the mean values and standard deviations. The normality of the distributions was confirmed using the Shapiro–Wilk test. Variables from the preceding night and EEG and EMG during napping were normally distributed.

The variables from the night before the experiment and the results of EEG and EMG compared between the two bedding conditions were analyzed with a paired t-test. In addition, effect size statistic (d) was analysed to determine the magnitude of the effect independent of sample size (Dankel & Loenneke, 2021). Differences were interpreted using Cohen’s (d) guidelines as trivial (<0.2), small (0.2–0.6), moderate (0.6–1.2), large (1.2–2.0), very large (2.0–4.0), and huge (>4.0) (Hopkins et al., 2009). To assess HRV between wakefulness and sleep, a two-way ANOVA (Napping condition [BC or UR] × Sleep status [Wakefulness or Sleep]) with repeated measures was performed for each variable. All calculations were performed using SPSS statistics version 28.0 (IBM SPSS Statistics for Windows; IBM Corp., Armonk, NY, USA).

Sleep and related parameters: electroencephalogram analysis

Sleep variables from assessments the night prior demonstrated no statistical difference between the two bedding conditions, showing similar results in each condition of the study (Table 1). Actigraphy data of four participants were excluded due to poor recording. The EEG data of one participant were unamenable to analyses because of poor recording as a result of unstable electrode placement. Table 2 presents sleep variables during napping, alongside nap-related parameters. With regard to sleep architecture, no significant differences were found between the two bedding conditions. Seven and six participants demonstrated N3 stage sleep in the BC and UR conditions, respectively. Three participants demonstrated REM sleep in both conditions. In addition, a CFF revealed no significant difference between the two bedding conditions (t = –0.056, df = 13, p = 0.779, d = 0.049). A spectral EEG analysis during each sleep stage showed no significant differences in any frequency band between the two bedding conditions (Table S1).

Electromyogram analysis during napping

Figure 3 shows the difference in surface EMG activity during sleep in each designated muscle alongside the grand mean of the two muscles. The EMG data of one participant were excluded from the analysis owing to poor EMG recording. The trapezius muscle demonstrated significantly lower EMG activity in the BC condition than in the UR condition (t = –2.276, df = 12, p = 0.019, d = 0.764). The external oblique muscle did not yield a significant difference in EMG activity between the BC and UR conditions (t = –1.493, df = 12, p = 0.093, d = 0.464). The grand mean of EMG activity across participants in the BC condition was significantly lower than that in the UR condition (t = –2.807, df = 12, p = 0.028, d = 0.693).

Heart rate variability analysis during napping and pre-nap wakefulness

Table 3 shows the HRV variables for each bedding condition between pre-nap wakefulness and NREM sleep during the nap experiment. Similar to the EMG recordings, the HRV data of three participants were excluded from the study owing to poor recording. The panels of Fig. 4 (one panel for each HRV variable) demonstrate the average values for each bedding condition recorded in pre-nap wakefulness and NREM sleep during napping. For all variables, there were no significant interactions between bedding condition and level of awareness. We found a significant main effect of bedding type for LF/HF (F[1, 20] = 4.314, p = 0.037, η_p² = 0.375), representing a lower LF/HF value in the BC condition than in the UR condition. There were no significant main effects in other HRV-related valuables associated with the level of alertness.