Abstract
Background
Sleep deprivation leads to decreased performance, alertness and degradation in the health status of a person. Often the person remains unaware of the reduced alertness and may end up taking inaccurate decisions. There was a need to study the sleep duration of college goers and to study the effect of total night-time sleep duration on daytime Electroencephalogram (EEG) characteristics.
Methods
A total of 30 asymptomatic volunteers were enrolled in a cross-sectional study. A baseline sleep diary for fourteen days was taken and one daytime recording was done while carrying out their normal daytime activity. EEG, electrooculography (EOG) and electromyography (EMG) were recorded, which were used for the scoring of their sleep-wake using the American Academy of Sleep Medicine criteria. EEG data were analysed using fast Fourier transform. Fast Fourier transform was done for sleep and wake durations using DOMINO software (©SOMNOMEDICS, Germany). A Python library was used to calculate power spectral density (PSD) in EEG frequency bands.
Results
A beta band was significantly reduced in the first 45 min of the recording as compared to the baseline. K means cluster showed clustering of two groups with a significant reduction of daytime sleep delta PSD values with reduced total sleep time and time in bed the previous night. The study concluded that subjects with reduced night-time sleep duration generated significantly lesser sleep delta PSD values.
Conclusion
Reduced duration of night-time sleep was associated with worse nap time sleep delta PSD values and vice versa. Interestingly, good quality sleep at night correlated better with delta PSD values of daytime recording.
Keywords: Sleep delta, Nighttime sleep duration, Daytime sleepiness, Daytime EEG, Nap
Introduction
In the modern world, students do not provide themselves with adequate sleep duration. There is a prevalence of significant sleep restriction among them. Various studies are conducted to find a link between their daytime efficiency and nocturnal sleep duration.1
In the normal day-to-day life, a lot of college students have restricted sleep schedules, and are also unable to maintain good sleep hygiene. Due to their busy lifestyles, they do not give themselves enough time for sleep. This sleep insufficiency is likely to cause daytime effects of increased sleepiness. As a student, there is always a denial and often the person is unaware of microsleep or increased sleepiness, which might creep in during daytime activity as a result of inadequate sleep.2,3
Sleep deprivation leads to decreased performance, alertness and degradation in the health status of a person.4 There was a need to study the sleep duration of college goers and to study the effect of sleep duration on daytime Electroencephalogram (EEG) characteristics.
Data regarding night-time sleep duration and on naps have been studied5 and their effect on daytime EEG is limited, and so our research question was–Does night-time sleep duration affect EEG spectra during the daytime?
Material and methods
A total of 30 asymptomatic college student volunteers in the age group of 18–25 years participated in the study after obtaining informed consent. The study was carried out at a medical college, and Institutional Ethical Clearance was duly taken. Local advertisements in the form of posters at college festivals all over the city were done for inviting volunteers.
The exclusion criteria were diagnosed cases of insomnia, narcolepsy, sleep apnoea and other sleep disorders and known cases of systemic disorders like hypertension, coronary artery disease, chronic obstructive pulmonary disease, bronchial asthma, epileptiform disorders, neuromuscular junction disorders and strabismus. Cases that had irregular sleep-wake schedules were excluded from the study as well.
Thirty asymptomatic individuals, who conformed to the exclusion and inclusion criteria, were selected. A detailed medical history was taken to rule out systemic and sleep-associated co-morbidities. Every individual completed a baseline (BL) 14 days sleep diary to check for a regular sleep-wake pattern. Since the comparison of night-time sleep duration and daytime parameters of the same individual was done, sleep diary was recorded in detail including caffeine intake, alcohol, smoking, use of electronic media, etc., and each individual was instructed to be kept similar on all days.
For daytime recording, a portable polysomnography machine (©SOMNOMEDICS, Germany) was placed on the subject on any one working day during the afternoon. EEG, electrooculography (EOG) and electromyography (EMG) were recorded for 2 h, which were used for the scoring of their sleep-wake duration using the American Academy of Sleep Medicine (AASM) criteria.6 EEG data were analysed using fast Fourier transform (FFT). FFT was done for sleep and wake durations using DOMINO software (©SOMNOMEDICS, Germany). Details of hook up: EEG was done using reference electrodes and the placement at F3, C3 and O1 was referenced to M2. Along with 2 channels of electrooculography (EOG) on each side of both the eyes, electromyography (EMG) electrodes on the chin were placed using gold plated electrodes (©NATUS, USA). The application of electrodes was done using a 10–20 system for EEG and AASM criteria for EEG, EOG and EMG. The surface of the skin was cleaned and electrodes were placed. The impedance of electrodes was kept <5 k-ohm.
Data Analysis: Data were analysed in DOMINO software (©SOMNOMEDICS, Germany). The parameters of daytime recording were—total recording time (TRT), total sleep time (TST), sleep latency (SL) for the onset of sleep and SLs, SL N1, SL N2, SL N3 (for N1, N2 and N3, respectively) and for rapid eye movement (REM) sleep. Duration of wake and sleep with different stages—N1, N2, N3 and REM, if any, were calculated. The FFT values of sleep time and wake time during the recording time were calculated in (μV2) and (%) of the total for (alpha + beta) frequency, sigma frequency and delta frequency.
The sleep diary (sd) parameters of the night before the recording were sdTIB (total time spent in bed during night), sdTST (total sleeping time overnight), sdWASO (wake after sleep onset), i.e. the amount of wake duration at night after the patient goes off to sleep, i.e. sleep breaks duration, sdSOL (sleep onset latency), i.e. time duration between going to bed and sleeping and sdSQ (sleep quality given by the subjects according to their subjective experience—Sleep quality was entered by the subject’s rating one's sleep subjectively on a scale of 0–10, where ‘0’ was very bad sleep and‘10’ was the best sleep possible.) and sdNAPS (naps duration taken during the day before the day of experiment).
Pearson correlation of the sleep diary parameters of the night, i.e. sdTIB, sdTST, sdWASO, sdSOL, sdSQ and sdNAPS (day before the experiment), with daytime recording EEG characteristics was done. If the correlation between two variables was greater than 0.20 or lower than −0.20, they were considered for checking the significance of the relationship. To check the significance of relationship between them, we performed a simple linear regression analysis and tested for significance of the regression model. A K-means cluster analysis was done to check for clusters of comparable spatial extent using the sleep parameters and FFT analysis. The results obtained were analysed to draw conclusions about the relation of sleep duration/quality on the daytime EEG recording. An overview of the study design is shown in Fig. 1.
Fig. 1.
The study design overview.
Daytime EEG data were manually checked, and artefact free segments of the awake BL and 30 s data just before the completion of the first 45 min (to measure sleepiness) were taken in case the sleep-wake analysis showed a wake state in these 45 min. In case there were episodes of sleepiness, the period where an individual felt asleep in the first 45 min was picked up and put for a power spectral density (PSD) analysis.7 In the second case, the data of sleep time were taken for the first 30 s, starting from the sleep episode. Both of these for the analysis and results purposes have been referred to as ‘first 45 min’ since it was to measure the sleepiness. A spectral analysis of the EEG signal was performed using an open source MNE-Python library, which is a python-based library.8 The spectral analysis approach incorporated the following steps: a) loading data, b) pre-Processing, c) segmenting the continuous data on to epochs, d) a time-Frequency analysis on the epoch EEG data and e) extracting the EEG spectral features.
The initial step included loading the raw EEG data, and then the pre-processing was performed on the loaded raw EEG signal. The pre-processing of EEG signal included filtering, artefacts detection, repair and custom re-referencing. The raw EEG data was filtered through setting the low frequency with 1 Hz (low pass) and the high frequency with 60 Hz (high pass) and the notch filtering performed by removing the notch frequency at 50 Hz. The EEG signal included bad spans of data, the eye movements and the blinks artefacts. The annotated bad segments were excluded using the MNE-Python library, which incorporated a technique, which automatically drops the bad segments from the EEG signal analysis at the time of the epochs creation from the continual data. The eye movements and blinks were repaired using a regression technique. On the obtained cleaned EEG signal, segmentation was applied in order to segment the continuous data on to epochs. A time-frequency analysis on the epoch EEG data was performed through the computation of the PSD of the EEG signal using Welch's method. The decomposition of the EEG signal into frequency bands was achieved using Fourier transforms (FFT). Ranges of the frequency bands were as follows: Delta (1–4 Hz), Theta1 (4–6 Hz), Theta2 (6–8 Hz), Alpha1 (8–10 Hz), Alpha2 (10–12.5 Hz) and Beta (13–30 Hz). Relative band power and ratios between the frequency bands were used as features in the analysis of EEG data. The extraction of the features was performed on the selected segments of data with a 512 ms window time and a 126 Hz sampling rate. The mentioned spectral analysis approach was used in the feature extraction process. K-means clustering was used, which is an unsupervised non-linear algorithm that clusters data based on similarity. An analysis was conducted in R-software version 4.1.2, and clustering was based on sdTST and sleep delta PSD of the daytime EEG to form two clusters. All guidelines as per the Declaration of Helsinki and good clinical practice guidelines were followed.
Results
Thirty volunteers for the study during the month of May and June 2021 were recruited after obtaining informed consent. Due to COVID restrictions during the second wave and subjects contracting COVID, only 22 subjects could complete the study. The daytime recording using PSG with sleep-wake characteristics, TRT, TST, wake duration and sleep efficiency (out of 100) in mean ± standard deviation (SD) is shown in Fig. 2.
Fig. 2.
The daytime recording of polysomnographic sleep-wake characteristics (total recording time, total sleep time, wake duration and sleep efficiency [(total sleep time/total recording time)∗100] of the subjects in mean ± standard deviation.
A daytime polysomnography analysis for sleep-wake parameters was done using AASM criteria, and Fig. 2, Fig. 3 show the sleep-wake characteristics and duration of various stages, respectively. The subjects were told to write the sleep diary and these parameters of the night before the day of the experiment, i.e. sdTIB, sdTST, sdWASO, sdSOL and sdNAPS, which are shown in Fig. 4. The mean ± SD value of sdSQ was 7.14 ± 1.61.
Fig. 3.
Daytime sleep-wake durations during the recording using polysomnography of subjects.
Fig. 4.
Sleep diary parameters of subject's night before the daytime recording.
The daytime recording SL was calculated for sleep and also for each stage, which could be for N1, N2, N3 or REM SL and N3 SL (Deep SL), and these are shown in Fig. 5. The mean ± SD of N1 latency and N2 latency were (37.63 ± 45.13) min and (22.94 ± 30.22) min, respectively. The mean ± SD of N3 latency was (3.27 ± 8.55) and mean ± SD of REM latency was (5.93 ± 16.64), subsequently.
Fig. 5.
Sleep latency of subjects during the daytime recording of sleep and for REM sleep (REM latency) and N3 Sleep (deep sleep latency) in minutes.
Data from the daytime polysomnography record were categorised into sleep and wake. The EEG recordings of the sleep and wake duration were then analysed for FFT. Values of different frequencies, (alpha + beta), sigma and delta, are depicted for sleep and wake duration in Fig. 6.
Fig. 6.
The above bar graph shows the distribution of mean ± standard deviation of the fast Fourier transform values of different frequencies [i.e. (alpha + beta), sigma and delta] during wake and sleep durations in the daytime recording.
The average sleep time during the day recording was (29.16 ± 24.67) min out of the average TRT for subjects, which was(123.09 ± 31.06) min. The subjects were found to be even in N3 sleep at times. The average N3 duration was (3.27 ± 8.55) min.
The relationship between daytime sleep parameters and FFT power values is given in Table 1.
Table 1.
Correlation between daytime recording sleep parameters & fast Fourier transformpower values.
| Sleep Parameters During Daytime Recording (Dependent Variable) | FFT Power Values (Independent Variable) | Correlation | p-Value |
|---|---|---|---|
| Sleep (min) | FFT Wake Alpha + Beta [μv2] | 0.4898 | 0.0207a |
| FFT Wake Sigma [μv2] | 0.4403 | 0.0403a | |
| FFTWake Delta [μv2] | 0.4936 | 0.0196a | |
| FFT Wake Alpha + Beta (%) | −0.2333 | 0.2961 | |
| FFT Wake Sigma (%) | −0.2095 | 0.3493 | |
| FFT Wake Delta (%) | 0.2481 | 0.2656 | |
| FFT Sleep Alpha + Beta [μv2] | 0.3891 | 0.0734 | |
| FFT Sleep Sigma [μv2] | 0.5576 | 0.0070a | |
| FFT Sleep Delta [μv2] | 0.5861 | 0.0042a | |
| FFT Sleep Sigma (%) | 0.2021 | 0.3662 | |
| FFT Sleep Delta (%) | 0.5823 | 0.0045a | |
| Wake (min) | FFT Wake Delta [μv2] | −0.2727 | 0.2196 |
| FFT Wake Alpha + Beta (%) | 0.3076 | 0.1637 | |
| FFT Wake Delta (%) | −0.3012 | 0.1731 | |
| FFT Sleep Alpha + Beta [μv2] | −0.3365 | 0.1257 | |
| FFT Sleep Sigma [μv2] | −0.4901 | 0.0206a | |
| FFT Sleep Delta [μv2] | −0.5605 | 0.0067a | |
| FFT Sleep Delta (%) | −0.2950 | 0.1826 | |
| REM (min) | FFT Wake Alpha + Beta [μv2] | 0.5808 | 0.0046a |
| FFT Wake Sigma [μv2] | 0.5703 | 0.0056a | |
| FFT Wake Delta [μv2] | 0.6917 | 0.0004a | |
| FFT Wake Alpha + Beta (%) | −0.4433 | 0.0388a | |
| FFT Wake Sigma (%) | −0.3767 | 0.0840 | |
| FFT Wake Delta (%) | 0.3936 | 0.0699 | |
| FFT Sleep Delta [μv2] | 0.2480 | 0.2657 | |
| FFT Sleep Delta (%) | 0.4076 | 0.0597 | |
| N1 Sleep (min) | FFT Sleep Sigma [μv2] | 0.2930 | 0.1858 |
| FFT Sleep Sigma (%) | 0.4523 | 0.0345a | |
| FFT Sleep Delta (%) | 0.2228 | 0.3190 | |
| N2 Sleep (min) | FFT Wake Alpha + Beta [μv2] | 0.2957 | 0.1816 |
| FFT Wake Sigma [μv2] | 0.2718 | 0.2211 | |
| FFT Wake Delta [μv2] | 0.2681 | 0.2276 | |
| FFT Sleep Alpha + Beta [μv2] | 0.2803 | 0.2064 | |
| FFT Sleep Sigma [μv2] | 0.5603 | 0.0067a | |
| FFT Sleep Delta [μv2] | 0.5260 | 0.0119a | |
| FFT Sleep Sigma (%) | 0.2195 | 0.3263 | |
| FFT Sleep Delta (%) | 0.4236 | 0.0495a | |
| N3 Sleep (min) | FFT Sleep Alpha + Beta [μv2] | 0.3606 | 0.0992 |
| FFT Sleep Sigma [μv2] | 0.6126 | 0.0024a | |
| FFT Sleep Delta [μv2] | 0.6610 | 0.0008a | |
| FFT Sleep Delta (%) | 0.3069 | 0.1648 | |
| Sleep Latency [m] | FFT Sleep Alpha + Beta [μv2] | 0.2827 | 0.2025 |
| FFT Sleep Alpha + Beta (%) | 0.5124 | 0.0148a | |
| FFT Sleep Sigma (%) | 0.2868 | 0.1956 | |
| FFT Sleep Delta (%) | 0.2269 | 0.3099 | |
| Sleep Efficiency (%) | FFT Wake Alpha + Beta [μv2] | 0.3756 | 0.0849 |
| FFT Wake Sigma [μv2] | 0.3111 | 0.1588 | |
| FFT Wake Delta [μv2] | 0.3574 | 0.1025 | |
| FFT Sleep Alpha + Beta [μv2] | 0.4477 | 0.0367a | |
| FFT Sleep Sigma [μv2] | 0.6809 | 0.0005a | |
| FFT Sleep Delta [μv2] | 0.6010 | 0.0031a | |
| FFT Sleep Sigma (%) | 0.2961 | 0.1809 | |
| FFT Sleep Delta (%) | 0.4928 | 0.0198a |
p-values <0.05.
Last night sleep diary parameters were correlated with various daytime EEG and PSG characteristics. These parameters with their Pearson correlation values and p-values for the significance of the relationship based on the significance of linear regression model test are shown in Table 2. The table summarizing the power spectrum density values of the BL and artefact-free segment in the first 45 min of PSG recording is shown in Table 3, and the ratio of PSD values in various frequency bands is given in Table 4, subsequently.
Table 2.
Correlation analysis of sleep diary parameters with daytime polysomnography recording.
| SD Parameter (Dependent variable) | Independent variable | Correlation | p-Value |
|---|---|---|---|
| sdTIB(min) | TRT (min) | 0.3112 | 0.1586 |
| Wake (min) | 0.2095 | 0.3494 | |
| REM (min) | 0.3622 | 0.0976 | |
| FFT Wake Sigma [μv2] | 0.2110 | 0.3459 | |
| Wake Delta (%) | −0.3663 | 0.0936 | |
| FFT Sleep Alpha + Beta [μv2] | 0.1961 | 0.3819 | |
| Sleep Alpha + Beta (%) | 0.2407 | 0.2805 | |
| sdTST (min) | TRT (min) | 0.3387 | 0.1231 |
| Wake (min) | 0.2116 | 0.3445 | |
| REM (min) | 0.3958 | 0.0682 | |
| FFT Wake Sigma [μv2] | 0.2003 | 0.3715 | |
| FFT Wake Delta (%) | −0.3590 | 0.1008 | |
| FFT Sleep Alpha + Beta (%) | 0.2192 | 0.3271 | |
| sdSOL (min) | N2 (min) | 0.2797 | 0.2075 |
| N3 (min) | 0.3817 | 0.0796 | |
| FFT Wake Alpha + Beta [μv2] | 0.2024 | 0.3663 | |
| FFT Wake Alpha + Beta (%) | 0.2305 | 0.3020 | |
| FFT Wake Sigma (%) | 0.3262 | 0.1384 | |
| FFT Wake Delta (%) | −0.2322 | 0.2983 | |
| Sleep Alpha + Beta [μv2] | 0.2575 | 0.2472 | |
| FFT Sleep Sigma [μv2] | 0.2111 | 0.3458 | |
| sdWASO(min) | Naps (min) | −0.3013 | 0.1730 |
| N2 (min) | −0.2994 | 0.1759 | |
| Sleep efficiency (%) | −0.2498 | 0.2623 | |
| SL [m] | 0.3983 | 0.0664 | |
| FFT Wake sigma (%) | −0.2293 | 0.3046 | |
| sdSQ | TRT (min) | 0.2242 | 0.3159 |
| FFT Wake Alpha + Beta (%) | 0.2977 | 0.1784 | |
| FFT Wake Delta (%) | −0.2190 | 0.3275 | |
| FFT Sleep Alpha + Beta [μv2] | 0.3120 | 0.1575 | |
| FFT Sleep Sigma [μv2] | 0.3261 | 0.1385 | |
| FFT Sleep Alpha + Beta (%) | 0.3588 | 0.1011 | |
| FFT Sleep Sigma (%) | 0.3680 | 0.0920 | |
| FFT Sleep delta (%) | 0.2074 | 0.3544 | |
| sdNAPS LAST DAY (min) | TRT (min) | −0.2161 | 0.3341 |
| REM (min) | −0.2099 | 0.3485 | |
| N1 (min) | 0.3517 | 0.1085 | |
| Wake Alpha + Beta (%) | 0.3442 | 0.1168 | |
| FFT Wake sigma (%) | 0.2284 | 0.3066 | |
| FFT Sleep Sigma (%) | 0.2198 | 0.3257 | |
| sdNAPS(min)+sdTST (min) | REM (min) | 0.2491 | 0.2637 |
| FFT Wake Alpha + Beta (%) | 0.3828 | 0.0787 | |
| FFT Wake Sigma (%) | 0.2724 | 0.2201 | |
| FFT Wake Delta (%) | −0.4180 | 0.0529 | |
| Sleep Alpha + Beta [μv2] | 0.2082 | 0.3526 | |
| FFT Sleep Alpha + Beta (%) | 0.3263 | 0.1383 | |
| FFT Sleep Sigma (%) | 0.3298 | 0.1339 |
Table 3.
Power spectral density (in dB) of the baseline and first 45 min of the test.
| Sl. no. | Relative Band Power - Power spectral Density |
|||
|---|---|---|---|---|
| Frequency Band | First 45 min Test (Mean ± SD) | Baseline (Mean ± SD) | t-Test (p-value) | |
| 1 | Delta (1–4 Hz) | 0.2101 ± 0.0062 | 0.1934 ± 0.0027 | 0.4984 |
| 2 | Theta1 (4–6 Hz) | 0.1318 ± 0.0015 | 0.1311 ± 0.0017 | 0.9597 |
| 3 | Theta2 (6–8 Hz) | 0.0969 ± 0.0010 | 0.0927 ± 0.0013 | 0.7003 |
| 4 | Alpha1 (8–10 Hz) | 0.0986 ± 0.0040 | 0.0867 ± 0.0024 | 0.565 |
| 5 | Alpha2 (10–12.5 Hz) | 0.0739 ± 0.0039 | 0.0723 ± 0.0024 | 0.9318 |
| 6 | Beta(13–30 Hz) | 0.1274 ± 0.0036 | 0.1837 ± 0.0066 | 0.032 |
Table 4.
Ratio between the power spectral density of different bands.
| Sl. No. | Ratio Between Frequency bands |
|||
|---|---|---|---|---|
| Ratio | First 45 min Test (Mean ± SD) | Baseline (Mean ± SD) | t-Test (p-value) | |
| 1 | Delta/Beta | 4.8738 ± 112.7615 | 1.3617 ± 1.1281 | 0.1864 |
| 2 | Theta1/Beta | 2.1277 ± 9.9850 | 0.9602 ± 0.6692 | 0.1571 |
| 3 | Theta2/Beta | 1.1509 ± 1.1325 | 0.6832 ± 0.3136 | 0.1025 |
| 4 | Theta1/Alpha1 | 2.2254 ± 2.8935 | 1.9013 ± 1.2134 | 0.5075 |
| 5 | Theta2/Alpha1 | 1.2694 ± 0.3167 | 1.2078 ± 0.2492 | 0.7165 |
| 6 | Theta1/Alpha2 | 3.2117 ± 7.9774 | 2.6014 ± 5.2783 | 0.4842 |
| 7 | Theta2/Alpha2 | 1.8027 ± 1.0290 | 1.6608 ± 0.0978 | 0.6941 |
| 8 | Delta/Theta1 | 1.6279 ± 0.3125 | 1.5192 ± 0.0978 | 0.4369 |
| 9 | Delta/Theta2 | 2.7100 ± 4.6404 | 2.3473 ± 0.9197 | 0.5228 |
| 10 | Delta/Alpha1 | 4.0863 ± 19.0010 | 2.8519 ± 2.5451 | 0.2924 |
| 11 | Delta/Alpha2 | 6.1793 ± 70.2302 | 3.7945 ± 9.5215 | 0.2913 |
K means clustering formed two distinct groups, as shown in Fig. 7. The details of the two cluster groups are mentioned in Table 5. A T-test indicated that there was a significant change between the two clusters, and the two clusters showed sdTIB in min (422.50 ± 70.76, 330 ± 41.97; p = 0.0013), sdTST in min (411.08 ± 70.71, 317 ± 36.44; p = 0.0009) and sleep delta (μV2) (41.92 ± 19.34, 19.6 ± 12.01; p = 0.0038).
Fig. 7.
K-means cluster based on total sleeping time overnight and daytime sleep delta power spectral values (the mean of each cluster is marked as cross in the same colour as the cluster group colour; cluster group 1 in red and cluster group 2 in black).
Table 5.
Details of cluster groups using k-means clustering.
| Parameter | Cluster1 |
Cluster2 |
t-Test |
|---|---|---|---|
| Mean ± SD | Mean ± SD | (p-value) | |
| sdSQ | 6.6 ± 1.96 | 7.58 ± 1.16 | 0.1839 |
| sdSO | 8.0 ± 6.65 | 8.50 ± 8.62 | 0.8796 |
| sdWASO | 5.0 ± 10.00 | 2.92 ± 8.65 | 0.6114 |
| sdTIB | 330.0 ± 41.97 | 422.50 ± 70.76 | 0.0013 |
| sdNaps | 41.5 ± 65.91 | 15.00 ± 37.29 | 0.2781 |
| sdTST | 317.0 ± 36.44 | 411.08 ± 70.71 | 0.0009 |
| Daytime sleep Delta [μv2] | 19.6 ± 12.01 | 41.92 ± 19.34 | 0.0038 |
Discussion
Sleep deprivation or sleep restriction is very prevalent among college going students. Mah et al. showed a sleep restriction of less than 7 h in an estimated 39.2% of college students, and excessive daytime sleepiness was found to be associated with the sleep duration of an average of 7 h.9,10 In our study, we found that the average time spent by the subjects for sleeping at night was (368.32 ± 74.07) min. The time spent in bed was also low (380.45 ± 74.83) min. This meant that the opportunity given by these college going students to sleep at night time was low, implying sleep restriction in our subjects.
Despite a heavy college schedule, students were taking daytime naps, and the average duration of nap was (27.05 ± 52.66) min. In a study, Ong et al. observed that naps help to decrease sleep restriction without compromising the normal sleep architecture.5 Longer naps greater than 30 min had impairment from sleep inertia for a short time but had an increased cognitive performance for a very long duration.11 It may be possible that our subjects were sleeping almost 6 h at night, and thus naps were probably a way of overcoming excessive sleepiness. Daytime naps were found with duration of even 7 h in college goers.10
Sleep deprived students remain sleepy during daytime and have impaired attention but are often unaware. We took afternoon time during college hours to study their EEG. Subjects were found to have sleep episodes during the afternoon recording.
The question whether the sleep parameters of the last day affect the sleepiness the next day can be answered by correlating the sleep parameters with the daytime EEG characteristics. We found cluster classification based on the night-time sleep duration and daytime sleep delta, where there was a significant change in sdTIB. This implies that a subject with better night-time sleep duration was able to generate better quality daytime sleep delta.
Also, extensive naps of duration more than 45 min are found to be an unhealthy practice and sets in a vicious cycle of affecting the sleep debt.7 Naps of less than 30 min are considered beneficial, and naps as short as of 10 min have shown promising effect.7,12
Another question which comes to the mind is that should individuals avoid being sleep deprived? Can good sleepers become better in taking naps or microsleep and when can being sleep deprived affect their objective quality of microsleep or daytime sleep? There is a need to educate college goers about good sleep hygiene practices; avoiding distracters close to bed time like electronic gadgets is important. The relative prevalence of sleep restriction in our study highlights that the need of sleep awareness and sleep education programs for college goers exists, which has been documented to be beneficial.13
Ong et al. conducted a study on young adults and assessed their sleep cycle, with restrictions involving nap and without nap sessions. A total of 57 adolescents were recruited for the two phases of sleep restriction (5-h TIB) and recovery (9-h TIB). Though naps had an impact on reducing sleep pressure, the variation in EEG could not be reduced by 9-h recovery sleep after a sleep restriction of 5 h. Hence, they concluded there was no substitute for nocturnal sleep. Also, recovery sleep did not help much, as non-REM duration increased and N2 latency decreased during the second sleep restriction after the first phase.5
Though we found significant correlation between WASO with REM latency while recording, WASO was only 3.86 ± 9.12 min. More data may be required to understand this observation.
In our study, it was clear that sleep restriction is very common in college goers and naps are the usual method to overcome this. If the quality of night sleep is reported better, they usually can restore their sleep by a good quality short nap or microsleep duration, the next day. This study highlights that daytime is affected by previous night's sleep duration. The effects of short sleep duration are deleterious as evidenced in our study, with the episodes of sleep and subtle alteration in awake EEG states. A good sleep time at night is important for a person to remain alert and not sleep during the daytime. The number of subjects is a limitation, and more studies on nap time and FFT parameters should be conducted to understand this correlation better. More EEG intense analysis on a larger group might give more direction.
Another imminent need which emerges is the need to educate our young generation about the subtle deterioration of daytime performance with shorter sleep duration at night-time. This is important in public interest to take care of good sleep habits in order to have an alert and productive population.14,15
Conclusion
College goers do not take enough opportunity to sleep. Reduced duration of night-time sleep was associated with worse nap time sleep delta PSD values and vice versa. Interestingly, good quality sleep at night correlated better with delta PSD of the daytime recording.
Patients/ Guardians/ Participants consent
Participants informed consent was obtained.
Ethical clearance
Institute/hospital ethical clearance certificate was obtained.
Source of support
The conduct of the study was funded as an ICMR STS project. Two of the authors who helped in the study were also funded as JRF and RA-1 position from the Department of Science and Technology (DST) SATYAM, India Scheme.
Disclosure of competing interest
The authors have none to declare.
Acknowledgements
The authors are thankful to all the subjects and sleep lab staff and the DST SATYAM scheme for the funding of JRF and RA-1 in the lab.
References
- 1.Yoon I.Y., Kripke D.F., Youngstedt S.D., Elliott J.A. Actigraphy suggests age-related differences in napping and nocturnal sleep. J Sleep Res. 2003;12(2):87–93. doi: 10.1046/j.1365-2869.2003.00345.x. [DOI] [PubMed] [Google Scholar]
- 2.Brown T.A., Chorpita B.F., Korotitsch W., et al. Psychometric properties of the depression anxiety stress scales (DASS) in clinical samples. Behav Res Ther. 1997;35:79–89. doi: 10.1016/s0005-7967(96)00068-x. [DOI] [PubMed] [Google Scholar]
- 3.Hershner S., Chervin R. Causes and consequences of sleepiness among college students. NSS. 2014;73 doi: 10.2147/NSS.S62907. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Killgore WDS, Weber M. Sleep deprivation and cognitive performance. In: Bianchi MT (ed) Sleep Deprivation and Disease. New York, NY: Springer New York, pp. 209–229.
- 5.Ong J.L., Lo J.C., Gooley J.J., et al. EEG Changes Accompanying Successive Cycles of Sleep Restriction with and without Naps in Adolescents. Sleep. 2017 doi: 10.1093/sleep/zsx030. Epub ahead of print 3 March. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Berry R.B., Brooks R., Gamaldo C.E., Harding S.M., Marcus C., Vaughn B.V. vol. 176. American Academy of Sleep Medicine; 2012. p. 2012.https://aasm.org/wp-content/uploads/2017/11/Summary-of-Updates-in-v2.0-FINAL.pdf (The AASM Manual for the Scoring of Sleep and Associated Events. Rules, Terminology and Technical Specifications, Darien, Illinois). [Google Scholar]
- 7.Dhand R., Sohal H. Good sleep, bad sleep! The role of daytime naps in healthy adults. Curr Opin Intern Med. 2007;6:91–94. doi: 10.1097/01.mcp.0000245703.92311.d0. [DOI] [PubMed] [Google Scholar]
- 8.Gramfort A., Luessi M., Larson E., et al. MNE software for processing MEG and EEG data. Neuroimage. 2014;86:446–460. doi: 10.1016/j.neuroimage.2013.10.027. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Mah C.D., Kezirian E.J., Marcello B.M., Dement W.C. Poor sleep quality and insufficient sleep of a collegiate student-athlete population. Sleep Health. 2018;4(3):251–257. doi: 10.1016/j.sleh.2018.02.005. [DOI] [PubMed] [Google Scholar]
- 10.Tsou M.T., Chang B.C. Association of depression and excessive daytime sleepiness among sleep-deprived college freshmen in northern Taiwan. Int J Environ Res Publ Health. 2019 Jan;16(17):3148. doi: 10.3390/ijerph16173148. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Lovato N., Lack L. The effects of napping on cognitive functioning. Prog Brain Res. 2010 Jan 1;185:155–166. doi: 10.1016/B978-0-444-53702-7.00009-9. [DOI] [PubMed] [Google Scholar]
- 12.Milner C.E., Cote K.A. Benefits of napping in healthy adults: impact of nap length, time of day, age, and experience with napping. J Sleep Res. 2009 Jun;18(2):272–281. doi: 10.1111/j.1365-2869.2008.00718.x. [DOI] [PubMed] [Google Scholar]
- 13.Alshahrani M., Al Turki Y. Sleep hygiene awareness: its relation to sleep quality among medical students in King Saud University, Riyadh, Saudi Arabia. J Fam Med Prim Care. 2019 Aug;8(8):2628. doi: 10.4103/jfmpc.jfmpc_359_19. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Hale L., Troxel W., Buysse D.J. Sleep health: an opportunity for public health to address health equity. Annu Rev Publ Health. 2020;41:81–99. doi: 10.1146/annurev-publhealth-040119-094412. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Dautovich N.D., MacPherson A.R., Ghose S.M., et al. Examining and promoting sleep health in the undergraduate classroom: a mixed-methods approach. Int J Environ Res Publ Health. 2021;18 doi: 10.3390/ijerph182312297. [DOI] [PMC free article] [PubMed] [Google Scholar]