Diurnal variations in the waking EEG: comparisons with sleep latencies and subjective alertness

Transcription

1 J. Sleep Res. (2000) 9, 243±248 Diurnal variations in the waking EEG: comparisons with sleep latencies and subjective alertness C. LAFRANCE and M. DUMONT Laboratoire de chronobiologie, Hoà pital du Sacre -Cúur de Montre al; De partement de Psychiatrie et Centre de recherche en sciences neurologiques, 1 Universite de Montre al, Que bec, Canada Accepted in revised form 27 January 2000; received 28 August 1999 SUMMARY Daytime measures of sleep latency and subjective alertness do not correlate with one another, suggesting that they assess di erent aspects of alertness. In addition, their typical diurnal variations show very di erent time courses. Quantitative analysis of the waking electroencephalogram (EEG) has been proposed as an objective measure of alertness, but it is not clear how it compares with other measures. In this study, the waking EEG was measured in the daytime to determine the presence of diurnal variations in the activity of standard frequency bands and to compare these variations with the temporal patterns typical of sleep propensity and subjective alertness. Alertness was evaluated in four men and 12 women, aged 19±33 y. Assessments were conducted every 2 h, from to 24.00, in the following order: a visual analogue scale of alertness, a waking EEG recording and a sleep latency test. The waking EEG was recorded with eyes open. For each recording session, 32±60 s of artefact-free signals were selected from the C3/A2 derivation, then subjected to amplitude spectral analysis. Four EEG frequency bands showed signi cant diurnal variations: delta, theta, sigma and beta1. None of these variations showed a signi cant correlation with the temporal patterns of sleep latencies or subjective alertness. At the individual level, however, theta band activity increased when subjective alertness decreased, suggesting that the theta band can be used to monitor variations in alertness in a given individual, even at the moderate levels of sleepiness experienced during the daytime. KEYWORDS alertness, MSLT, quantitative EEG, subjective alertness, waking EEG INTRODUCTION The measurement of alertness has become a matter of deep interest to those concerned with understanding its nature and the clinical treatment of sleep disorders. Both subjective and objective measures have been developed to evaluate changes in alertness. The Stanford Sleepiness Scale (SSS) and the visual analogue scale (VAS) are two of the most widely used subjective measures (Johnson 1992). These tests are easy and quick to administer and can be used in clinical settings, as well as in laboratory and eld research. However, being essentially subjective, these measures are in uenced by the subject's ability and motivation for introspection. Correspondence: Dr M. Dumont, Chronobiology Laboratory, Sacre - Coeur Hospital, 5400 blvd Gouin W., Montre al (Que bec), Canada H4J 1C5. Tel.: , ext. 2246; fax: ; M-Dumont@crhsc.umontreal.ca Among objective measures, the Multiple Sleep Latency Test (MSLT) is generally accepted as the `gold standard' for evaluating the physiological state of alertness in clinical and research settings (Carskadon et al. 1986). The test measures the physiological tendency to fall asleep and presupposes that the shorter the sleep latency, the lower the alertness level. A major problem with the MSLT, however, is that the propensity to fall asleep during the test does not necessarily imply that the subject is sleepy: it is not unusual to nd individuals with both normal levels of subjective alertness and very short sleep latencies on the MSLT (Harrison and Horne 1996). Sleep latency can also be a ected by the subject's motivation and state of mind. In the absence of sleep deprivation, levels of alertness vary across the daytime. Sleep latencies are shorter at midday and before bedtime, and are longer in the morning and late evening (Richardson et al. 1982). Numeric and analogue scales usually present stable levels of subjective alertness throughout the day Ó 2000 European Sleep Research Society 243

2 244 C. Lafrance and M. Dumont with a rapid decline around one's habitual bedtime (Richardson et al. 1982; Dijk et al. 1992). Thus, in contrast to sleep latencies, subjective measures do not show an afternoon decrease in alertness (Richardson et al. 1982; Dijk et al. 1992). It is not unusual for sleep latencies and subjective scores to be uncorrelated (Johnson et al. 1991). It has been suggested that the two measures evaluate di erent aspects or di erent types of daytime alertness, with di erent sensitivity for the various sources of sleepiness (e.g. sleep deprivation, circadian phase, context; Johnson 1992). Quantitative analysis of the waking EEG has also been proposed as an objective measure of sleepiness (Akerstedt and Gillberg 1990). Compared with the MSLT, this measure is more independent of the subject's motivation or ability to fall asleep and allows for a repeated estimation of the subject's level of alertness without the accumulation of sleep over the day. However, use of the waking EEG has not been validated as a standard measure of alertness. In fact, little is known about which of the waking EEG characteristics may represent a sign of physiological alertness or sleepiness. Most of the available information comes from sleep deprivation protocols. Sleep-deprived subjects show increased activity in the theta/alpha range of the EEG in association with an increase in subjective sleepiness (Akerstedt and Gillberg 1990; Cajochen et al. 1995; Aeschbach et al. 1997). This increase in the activity of slow frequency bands was considered to be the EEG expression of sleepiness which results from sleep deprivation and the cumulative hours of wakefulness. Parallel changes between theta/alpha frequencies and subjective sleepiness were also found in a study using daytime melatonin administration in non-sleep deprived subjects (Cajochen et al. 1996). A few studies have examined the time course of daytime waking EEG frequencies (Cacot et al. 1995; Lorenzo et al. 1995). Signi cant diurnal variations were found in each EEG band. In central derivations, maximum power in the EEG occurred at a later time for higher frequencies: the slow frequency bands peaked in the morning and in the afternoon, whereas the faster bands showed their maxima in the late afternoon and early evening. This pattern suggests that waking EEG activity is not stable throughout the day and that the observed variations di er across the various frequency bands. These data, however, were not compared with daytime variations in other measures of alertness such as subjective scales or sleep latencies. In this study, waking EEG was recorded from to bedtime in normal volunteers not subjected to sleep deprivation. The possible similarities between the variations observed in the EEG frequency bands and those found in subjective alertness and sleep latencies during the daytime were examined. Three speci c questions were considered: 1. In the absence of sleep deprivation, which of the waking EEG frequency bands show diurnal variations in normal subjects? 2. Are there some frequency bands which show the temporal patterns typical of sleep propensity or subjective alertness? 3. At the individual level, is there an association between the changes in the waking EEG theta/alpha frequency bands and the changes in the two measures of alertness during the daytime? METHODS Subjects Sixteen subjects (four men, 12 women) aged 19±33 y (mean SDˆ y) took part in the study. According to questionnaires and interview, they had no sleep or vigilance complaints and no history of psychiatric or neurological disorders. Except for eight women using hormonal contraceptives, subjects did not report using drugs or medications. Volunteers who had worked nights or travelled to another time zone in the past 3 months were excluded, as were subjects who had extreme scores of morningness or eveningness (Horne and Ostberg 1976). All subjects were nonsmokers and were required to abstain from alcohol and ca eine for the duration of the study. They signed a consent form approved by the institutional ethical committee and were paid for their participation. Procedures Prior to the laboratory part of the study, each subject was required to maintain a regular sleep schedule for four consecutive days, with bedtime around and waketime around Compliance was veri ed by sleep diaries and 24-h activity recordings. Alertness measures were administered during the daytime following a night of polysomnography in the laboratory. Throughout the testing day, the subjects had to remain in their room in dim illumination (< 30 lux) and were allowed to pursue quiet activities. Three measures were used to evaluate levels of alertness/ sleepiness during the day. The measures were conducted every 2 h, from to 24.00, in the following order: a 10-cm visual analogue scales of alertness (VAS), a waking EEG recording and a sleep latency test. Each VAS was a 10-cm line, with the inscriptions very sleepy on the left and very alert on the right. Subjects were instructed to draw a vertical bar across the line at the point corresponding to their subjective feeling of alertness at that time. Results were expressed in centimetres measured between the left end of the line and the bar drawn by the subject. Each VAS was administered immediately before the other two measures of vigilance. The EEG was recorded from four derivations (C3/A2, C4/ A1, O1/A2, O2/A1) in addition to the electromyogram (EMG) and the electrooculogram (EOG). Signals were recorded on paper with an eight-channel Grass polygraph (EEG sensitivity 7.5 lv/mm, bandpass 0.3±90 Hz) at a speed of 15 mm/s. The signals were also relayed to a personal computer where they were digitized at a sampling rate of 256 Hz and ltered with a digital lter having a cut-o frequency at 64 Hz. One of every two points was actually stored on disk (128 Hz).

3 Diurnal variations in the waking EEG 245 Daytime sleep latency tests were administered from to following standard procedures for the MSLT (Carskadon et al. 1986). The sleep latency recorded at was used as an eighth daytime sleep latency value. Except for the sleep latencies, subjects were awakened after sleep onset. Sleep latency was de ned as the time from lights o to the beginning of the rst minute of Stage 1 sleep, the rst epoch of another sleep stage, or as 20 min if no sleep occurred. Each test was scored in 20-s epochs according to standard criteria (Rechtscha en and Kales 1968) by two raters with an interrater reliability of more than 90%. When discrepancies occurred, decisions were made by consensus. Immediately before each sleep latency test, a waking EEG with eyes open was recorded. Each recording lasted for a maximum of 2 min during which the subjects had to keep their eyes open and xed on a target. Precise criteria for visual identi cation of artefacts were standardized within and between scorers. These criteria were devised to reject epochs contaminated by eye movements, muscular activity, sweating, cardiac signals and those suggesting Stage 1 sleep. For each recording session, 9±15 4-s epochs (32±60 s) were selected from the C3/A2 derivation. Of the 128 selections, 21 were shorter than 60 s and, on average, s (SEM ˆ 0.45) of recording were included in the analyses. Selected epochs were subjected to amplitude spectral analysis performed using commercial software (RHYTHM, Stellate Systems, Montreal, Canada) which calculated fast Fourier transform with a resolution of 0.25 Hz and cosine window smoothing. Absolute amplitudes (lv/hz) were added within six frequency bands: delta (0.75±3.75 Hz), theta (4.00±7.75 Hz), alpha (8.00± Hz), sigma (12.00±13.75 Hz), beta1 (14.00±19.75 Hz) and beta2 (20.00±31.00 Hz). Finally, the values obtained for each 4-s epoch were averaged across each recording session. Statistical analyses The rst set of analyses tested for temporal variations in subjective alertness, sleep latency and EEG activity among the eight times of day. For each subject, the results of each of the three measures obtained at each time of day were transformed into the percentage of the mean calculated over the eight sessions. One-way ANOVAs for repeated measures (with Huynh-Feldt corrections) were then conducted on the values. Signi cant ANOVAs were followed by post-hoc comparisons (Tukey HSD). A second set of analyses used correlations to compare quantitatively the temporal patterns between the EEG frequencies and the other two measures of alertness (Markowitz et al. 1988). The subjects scores (already transformed in percentage of the mean) were rst averaged for each time of testing. Spearman rank-order correlations were then used to compare the eight averaged values for each EEG frequency band and each of the other two measures of alertness. We used the nonparametric test because many correlations included non-normal distributions (mostly VAS scores). The last statistical analysis tested for a possible association between the three measures of alertness at an individual level. For this purpose, we chose the technique described by Monk et al. (1997), using a sign test applied on individual coe cients of correlation (Spearman rho) obtained for each pair of variables. Although this procedure could not test the strength of individual associations, it is useful to test whether two variables tend to show either a positive or negative correlation in a signi cant number of subjects. For the purpose of graphical representation, mean and standard error (SE) of rho were calculated across subjects for each pair of measures. RESULTS Diurnal variations A global ANOVA on measures of subjective alertness revealed a signi cant time-of-day e ect (F 7,105 ˆ 10.67, P < 0.001). The only signi cant di erence was between the low score obtained at and each of the other seven measures collected during the day (Tukey HSD: P < for all comparisons with the score at 24.00). There was no di erence between the measures of subjective alertness from to (Fig. 1). Daytime sleep latencies showed characteristic diurnal variations, with a decrease in sleep latencies at midday and an increase in the evening, followed by a shortening at bedtime (Fig. 2). The ANOVA was signi cant (F 7,105 ˆ 5.64, P < 0.001) and post-hoc comparisons revealed that the values at and were signi cantly higher than those obtained at 12.00, and h (P < 0.02 for each comparison). The decrease in sleep latencies between and was also signi cant (P < 0.04). Four EEG frequency bands had signi cant diurnal variations: delta (F 7,105 ˆ 2.83, P < 0.01), theta (F 7,105 ˆ 2.31, P < 0.03), sigma (F 7,105 ˆ 2.49, P < 0.02) and beta1 (F 7,105 ˆ 2.59, P < 0.02). The diurnal variations are illustrated in Fig. 3. The delta band showed high activity values at 16.00, which were signi cantly di erent to the low values obtained at (P < 0.01). The di erence between lowest and highest activity values was between and (P < 0.08) for the theta band, and between and (P < 0.01) for the beta1 band. For the sigma band, the low values at were signi cantly di erent to those reached at and (P < 0.05 for both comparisons). Variations in the alpha and beta2 bands were not statistically signi cant (P > 0.10). Group correlations between temporal patterns There was a negative correlation between variations in subjective alertness and the activity of the beta2 band (R ˆ ±0.83, P < 0.01). None of the other correlations calculated among the three measures of vigilance approached statistical signi cance (P > 0.10), including the correlation between variations in the subjective scores of alertness and those in sleep latencies (R ˆ ±0.45, P ˆ 0.26).

4 246 C. Lafrance and M. Dumont Figure 1. Time course of the VAS scores from to For each subject, VAS scores were transformed into the percentage of the mean calculated over the eight times of day. Each point represents the mean of 16 subjects ( SE). Intra-individual correlations between the three measures The sign test conducted on individual Spearman correlations revealed that, for most of the subjects (14/16), there was a negative correlation between the theta band and the subjective alertness scores (mean correlation: R ˆ ± ; range: 0.05 to ±0.55; sign test: P < 0.01). Twelve of the 16 subjects also showed a negative correlation which approached signi cance (P ˆ 0.08) between the subjective alertness scores and both the alpha and beta2 bands. None of the other pairs of variables showed a signi cant association (Fig. 4). DISCUSSION Subjective alertness, sleep latencies and four of the six EEG frequency bands displayed signi cant diurnal variations, but the three measures had di erent diurnal time courses. VAS and sleep latencies showed typical diurnal variations (Richardson et al. 1982; Dijk et al. 1992). Consistent with previous results (Cacot et al. 1995), higher frequencies of the waking EEG reached their maximum activity later during the day. Diurnal variations in beta2 activity were too small to be statistically signi cant but overall, the small increasing trend seen in the group data was closely correlated with the decreasing trend in subjective alertness. A similar association was reported in a previous study of waking EEGs recorded during 38-h of sleep deprivation (Dumont et al. 1999). Lorenzo et al. (1995) proposed that the increase in beta power (13.00±25.5 Hz) during sleep deprivation may result from e orts to stay awake. Although there was no sleep deprivation in this study, it is possible that the increase in beta2 activity re ected a mounting tension in the cranio-facial muscles while subjects' sleepiness increased (O'Donnell et al. 1974). At the individual level, 12 of the 16 subjects also showed a negative Figure 2. Time course of the sleep latencies from to For each subject, sleep latencies were transformed into the percentage of the mean calculated over the eight times of day. Each point represents the mean of 16 subjects ( SE).

5 Diurnal variations in the waking EEG 247 Figure 3. Time course of the amplitude of each EEG frequency band from to For each subject, EEG amplitudes were transformed into the percentage of the mean calculated over the eight times of day. Each point represents the mean of 16 subjects ( SE). correlation between the two measures, but this trend did not reach statistical signi cance (P ˆ 0.08). When calculated within subjects, the sign test revealed an inverse relationship between theta band activity and subjective alertness in most (14/16) of our subjects. This negative correlation is consistent with conclusions of protocols using sleep deprivation (Akerstedt and Gillberg 1990; Aeschbach et al. 1997). A similar association was found between VAS scores and the alpha band activity in 12 subjects, but the sign test was not signi cant (P ˆ 0.08). The alpha band seems less sensitive than the theta band to the changes in daytime alertness. In accordance with a previous study that used a pharmacological treatment to modify levels of alertness (Cajochen et al. 1996), our results suggest that increases in theta band activity may re ect increases in sleepiness per se, and are not just an indicator of sleep deprivation. The relationship between theta activity and subjective alertness did not persist when examined in group data. It seems that the association between the two measures was not strong enough to compensate for inter-individual variations when tested on group data. None of the changes in frequency bands of the waking EEG were correlated with variations in sleep latencies. This was more unexpected since both were physiological measures using EEG recordings. It should be noted that such a dissociation has also been reported between daytime sleep latencies and the subsequent sleep EEG (Dijk et al. 1987). However, contrary to the monotonic increase reported in the sleep EEG, the activity of the delta band in the waking EEG increased around the time of the shortest sleep latencies (16.00) but decreased when sleep latencies were at their longest (22.00) (Fig. 3). Therefore, it is possible that sleep propensity and activity of the slowest frequency band of the waking EEG are related to similar aspects of alertness. In contrast to others, we did not nd a signi cant diurnal variation in the alpha band. However, in one study (Cacot et al. 1995), the afternoon peak in the alpha band was due to only one subject who had a clear maximum in the 8-Hz frequency at In the other study (Lorenzo et al. 1995), half of the data were recorded during sleep deprivation, which may have increased the variations in alpha band activity. Studies that have sampled the waking EEG throughout the entire 24 h have shown signi cant changes in the alpha range Figure 4. Mean Spearman rho correlations ( SE) between sleep latencies (MSLT: open bars), subjective alertness scores (VAS: hatched bars) and the activity of each of the frequency bands of the waking EEG. Correlations were calculated intra-individually, then averaged over the 16 subjects for the illustration. Signi cance: *P < 0.05 (sign test on the 16 correlations).

6 248 C. Lafrance and M. Dumont primarily during the night and not during the daytime (Aeschbach et al. 1997; Dumont et al. 1999), which is consistent with our results. Previous reports of diurnal variations in the beta band included frequencies lower than 20 Hz (Cacot et al. 1995; Lorenzo et al. 1995). Considering the signi cant variations in the sigma (12.00±13.75 Hz) and beta1 (14.00± Hz) bands in this study, but not in the beta2 band (20.00±31.00 Hz), it is possible that the diurnal changes reported previously were due mostly to variations in frequencies lower than 20 Hz. The few studies that have recorded many EEG derivations have reported that the presence and magnitude of diurnal variations in EEG activities depend on the localization being examined (Cacot et al. 1995; Lorenzo et al. 1995). Consequently, recordings from other EEG derivations could reveal di erent time courses in diurnal variations as well as di erent associations with other measures of alertness. In conclusion, four of the six frequency bands in the waking EEG showed signi cant variations during the daytime. When group data were considered, none of these variations correlated with variations in subjective alertness or sleep propensity. However, our results suggest that the theta band could be used to monitor variations in alertness in a given individual, even at the moderate levels of sleepiness experienced during the daytime. Spectral analysis of the waking EEG is an increasingly accessible technique. More normative data, from di erent derivations and for various frequency ranges are required to exploit the potential of this new tool for the evaluation of alertness in humans. ACKNOWLEDGEMENTS We thank Dr Jean Paquet for his help with the analyses. This work was supported by a fellowship from the FCAR of Que bec (C.L.) and by a grant from the Medical Research Council of Canada (M.D.). REFERENCES Aeschbach, D., Matthews, J. R., Postolache, T. T., Jackson, M. A., Giesen, H. A. and Wehr, T. A. Dynamics of the human EEG during prolonged wakefulness: evidence for frequency-speci c circadian and homeostatic in uences. Neurosci. Lett., 1997, 239: 121±124. Akerstedt, T. and Gillberg, M. Subjective and objective sleepiness in the active individual. Int. J. Neurosci., 1990, 52: 29±37. Cacot, P., Tesolin, B. and Sebban, C. Diurnal variations of EEG power in healthy adults. Electroenceph. Clin. Neurophysiol., 1995, 94: 305±312. Cajochen, C., Brunner, D. P., Krauchi, K., Graw, P. and Wirz-Justice, A. Power density in theta/alpha frequencies of the waking EEG progressively increases during sustained wakefulness. Sleep, 1995, 18: 890±894. Cajochen, C., Krauchi, K., Von Arx, M. A., Mori, D., Graw, P. and Wirz-Justice, A. Daytime melatonin administration enhances sleepiness and theta/alpha activity in the waking EEG. Neurosci. Lett., 1996, 207: 1±5. Carskadon, M. A., Dement, W. C., Mitler, M. M., Roth, T., Westbrook, P. R. and Keenan, S. Guidelines for the multiple sleep latency test (MSLT): a standard measure of sleepiness. Sleep, 1986, 9: 519±524. Dijk, D. J., Beersma, D. G. M. and Daan, S. EEG power density during nap sleep: re ection of an hourglass measuring the duration of prior wakefulness. J. Biol. Rhythms, 1987, 2: 207±219. Dijk, D. J., Du y, J. F. and Czeisler, C. A. Circadian and sleep/wake dependent aspects of subjective alertness and cognitive performance. J. Sleep Res., 1992, 1: 112±117. Dumont, M., Macchi, M. M., Carrier, J., Lafrance, C. and He bert, M. Time course of narrow frequency bands in the waking EEG during sleep deprivation. Neuroreport, 1999, 10: 403±407. Harrison, Y. and Horne, J. A. `High sleepability without sleepiness'. The ability to fall asleep rapidly without other signs of sleepiness. Neurophysiol. Clin., 1996, 26: 15±20. Horne, J. A. and Ostberg, O. A self-assessment questionnaire to determine morningness±eveningness in human circadian rhythms. Int. J. Chronobiol., 1976, 4: 97±110. Johnson, L. C. Daytime sleepiness in good sleepers: measurement and correlates. In: R. J. Broughton and R. D. Ogilvie (Eds.) Sleep, Arousal, and Performance. Birkhauser, Boston, 1992: 220±229. Johnson, L. C., Freeman, C. R., Spinweber, C. L. and Gomez, S. A. Subjective and objective measures of sleepiness: e ect of benzodiazepine and ca eine on their relationship. Psychophysiology, 1991, 28: 65±71. Lorenzo, I., Ramos, J., Arce, C., Guevara, M. A. and Corsi-Cabrera, M. E ect of total sleep deprivation on reaction time and waking EEG activity in man. Sleep, 1995, 18: 346±354. Markowitz, M. E., Arnaud, S., Rosen, J. F., Thorpy, M. and Laximinarayan, S. Temporal interrelationships between the circadian rhythms of serum parathyroid hormone and calcium concentrations. J. Clin. Endocrinol. Metab., 1988, 67: 1068±1073. Monk, T. H., Buysse, D. J., Reynolds, C. F. III, Berga, S. L., Jarrett, D. B., Begley, A. E. and Kupfer, D. J. Circadian rhythms in human performance and mood under constant conditions. J. Sleep Res., 1997, 6: 9±18. O'Donnell, R. D., Berkhout, J. and Adey, W. R. Contamination of scalp EEG spectrum during contraction of cranio-facial muscles. Electroenceph. Clin. Neurophysiol., 1974, 37: 145±151. Rechtscha en, A. and Kales, A. A manual of standardized terminology, techniques and scoring system for sleep stages of human subjects. Public Health Service. US Government Printing O ce, Washington, DC, Richardson, G. S., Carskadon, M. A., Orav, E. J. and Dement, W. C. Circadian variation of sleep tendency in elderly and young adult subjects. Sleep, 1982, 5: S82±S94.