- Email: [email protected]
Body posture affects electroencephalographic activity and psychomotor vigilance task performance in sleep-deprived subjects
Clinical Neurophysiology 114 (2003) 23–31 www.elsevier.com/locate/clinph
Body posture affects electroencephalographic activity and psychomotor vigilance task performance in sleep-deprived subjects John A. Caldwell* ,1, Brian Prazinko, J. Lynn Caldwell 1 Aeromedical Research Laboratory, Fort Rucker, AL 36362-0577, USA Accepted 12 August 2002
Abstract Objective: This study examined the effects of posture on electroencephalographic (EEG) activity and psychomotor vigilance task (PVT) performance in 16 sleep-deprived volunteers. Methods: EEG data were collected while participants completed 10 min PVTs under two counterbalanced sitting/standing conditions during 28 h of continuous wakefulness. Results: In both the sitting and standing conditions, theta activity progressively increased as a function of sleep loss, but standing upright significantly attenuated this effect, suggesting that alertness was improved by the more upright posture. The PVT results showed that cognitive psychomotor performance was maintained at nearly well-rested levels by standing upright, whereas reaction time and attention noticeably deteriorated when participants were seated. Conclusions: These results suggest that an upright posture increases EEG arousal and sustained attention, indicating that postural manipulations can be useful for counteracting fatigue in sleep-deprived individuals. q 2002 Published by Elsevier Science Ireland Ltd. Keywords: Body posture; Electroencephalographic; Sleep deprivation; Psychomotor vigilance task; Performance
1. Introduction Although there is little direct evidence that posture influences the alertness of individuals who are fatigued, the scientific literature and anecdotal reports suggest it is easier to maintain wakefulness when sitting or standing than when lying down (Bonnet, 2000). Shift workers, who suffer from drowsiness on the job due to chronically shortened or disrupted sleep (Dinges, 1995), often state that standing up is one way of improving alertness at work, and it has been found that standing as opposed to lying down contributes to enhanced alertness, possibly by altering afferent sensory input (Dijkman et al., 1997). In fact, several studies have suggested that a more upright postural orientation actually inhibits sleepiness. Nicholson and Stone (1987) found that subjects experienced reductions in total sleep time, decreased sleep efficiency, and increased awakenings when they attempted to sleep in a more-upright sitting position (17.58 from the vertical angle), as opposed to either lying flat or reclining 49.58 or 378. Similar results * Corresponding author. Tel.: 11-210-536-8134. E-mail address: [email protected] (J.A. Caldwell). 1 Present address: U.S. Air Force Research Laboratory, 2485 Gillingham Drive, Suite 25, Brooks AFB, TX 78235, USA.
were reported by Aeschbach et al. (1994) who discovered that subjects who slept in reclining chairs (rather than lying flat in bed) experienced reduced sleep efficiency, less rapideye-movement (REM) sleep, and increased stage 1 sleep. The reasons for these postural effects are not completely understood, but changes in core body temperature, plasma catecholamine levels, and hemodynamic mechanisms may be involved. With regard to the role of body temperature, Kleitman and Doktorsky (1933) long ago reported that subjects manifested higher rectal temperatures when standing upright than when supine and vice versa, and these findings have been confirmed by Tikuisis and Ducharme (1996) and Kra¨uchi et al. (1997). Furthermore, changes in core body temperature have been associated with alterations in self-reported sleepiness – Kra¨uchi et al. (1997) found that higher core body temperatures were accompanied by lower self-ratings of tiredness (sleepiness) in subjects who were sitting/standing rather than lying down. With regard to the role of hemodynamic and/or catecholaminergic mediators, it is well known that standing up produces elevations in both heart rate and blood pressure which counteract the loss in cerebral blood flow stemming from an initial drop in venous pressure. Levels of urinary epinephrine and norepinephrine have been found to rise as a result of these hemodynamic changes (Sundin, 1956), and augmented adrenergic nerve
1388-2457/02/$ - see front matter q 2002 Published by Elsevier Science Ireland Ltd. PII: S 1388-245 7(02)00283-3
CLINPH 2002093
24
J.A. Caldwell et al. / Clinical Neurophysiology 114 (2003) 23–31
activity and stimulation of the adrenal medulla follows. Hickler et al. (1959) showed that homeostatically-driven elevations in heart rate are accompanied by a significant rise in the levels of plasma catecholamines. In addition to posturally-induced peripheral and neurendocrine effects, hemodynamic changes have been shown to influence central nervous system (CNS) activation. Cole (1989) found that heart rate and blood pressure elevations were accompanied by an increase in the amplitude-percentage product of high-frequency (beta) electroencephalographic (EEG) activity when subjects were tilted 408 upright on a tilt table. The upright tilt also inhibited sleepiness as evidenced by the fact that volunteers were able to stave off sleep for almost 6 min longer than when they were tested in the supine position. It was concluded that upright posture enhanced arousal via hemodynamically-induced reductions in baroreceptor firing (subjects were tested both with and without the application of positive leg pressure in an effort to isolate the effects of the baroreflex response). These findings suggest the importance of postural effects on sleepiness – a relationship which has been recognized by circadian rhythm/sleep deprivation researchers who have implemented ‘constant routine’ protocols in which subjects are prevented from standing and moving about while undergoing laboratory testing. However, the isolated impact of posture has not been well-studied in sleep-deprived individuals. Furthermore, the feasibility of using administrativelyimposed or self-imposed postural changes as a countermeasure for drowsiness in sleepy shift workers or other sleepdeprived personnel has not been established. In a previous study (Caldwell et al., 2000a), we found that requiring sleep-deprived volunteers to stand up, rather than remain seated, attenuated the fatigue-related elevations in both delta (1.5–3.0 Hz) and theta (3.0–8.0 Hz) activity that are commonly observed in sleepy individuals. We further found that this effect persisted throughout more than 55 h of continuous wakefulness. Although increased theta activity has been found to coincide with increased workload (Sterman et al., 1987; Wilson and Hankins, 1994) as well as increased fatigue, the widespread augmentation of theta activity has more traditionally been associated with impaired alertness and drug-induced sedation (Goldstein et al., 1968; Pigeau et al., 1987), especially in contexts where either of these factors were known to be present. In addition, these generalized low-frequency EEG patterns have been associated with decrements on vigilance tasks (Belyavin and Wright, 1987). Since this has been the case, we hypothesized that the apparent CNS stimulation offered by an upright postural intervention (Caldwell et al., 2000a) should also benefit cognitive psychomotor performance. However, no performance measures were collected in the earlier study. Because of this, the present investigation was performed in an effort to explore the effects of posture (sitting versus standing) both on EEG activity and concurrently-measured psychomotor vigilance task (PVT) performance. Specifically, it was of interest to learn whether the
apparent improvement in physiological arousal associated with standing (versus sitting) would translate into improved sustained attention. 2. Methods The data reported here were selected from a larger study in which the efficacy of temazepam was assessed for promoting daytime sleep in shift workers (further information is provided below). The entire study is explained in Caldwell et al. (2001). This research was approved by the U.S. Army Aeromedical Research Laboratory’s (USAARL’s) local Scientific and Human Use Committees prior to execution. In addition, the protocol was approved by the Human Subjects Review Board of the U.S. Army Medical Research and Materiel Command. 2.1. Subjects Fifteen males and one female resided at USAARL for a total of 10 days (although only 2 of these days are the focus of this report). The mean age of the volunteers was 33.7 years, with a range of 26–44 years. All subjects were medically evaluated prior to the study, and none were found to be experiencing any current significant illnesses. None of the participants had any evidence of past psychiatric problems or sleep disorders. Volunteers were not allowed to consume alcohol, caffeine, or any type of medication (other than acetaminophen and ibuprofen) during the protocol. Caffeine users were asked to eliminate caffeine consumption several days prior to the study. Only one subject used nicotine; he was allowed to smoke between test sessions, but not during testing. Given the time of the EEG sessions, the last cigarette before testing would have been no earlier than 45 min prior to the test. There were 3 left-handed subjects, and the remainder were right-handed. Each subject signed an informed consent agreement prior to their participation. 2.2. Apparatus 2.2.1. PVT The PVT (Ambulatory Monitoring, Inc., Ardsley, NY) is a portable, simple reaction time test known to be sensitive to sleep loss (Dinges et al., 1997). A 10 min trial of the PVT was used at each testing time in this investigation. The device visually displays numbers (milliseconds) that increment rapidly in sequence for up to 1.5 s until the volunteer responds by pressing a microswitch. The number of milliseconds from stimulus onset to response was recorded. At the outset, participants were admonished to respond to the stimulus as quickly as possible. If the participant responded later than 500 ms or failed to respond within 1.5 s, a ‘lapse’ was recorded. Since the time of this task was set for 10 min, the actual number of stimuli presented varied within each trial since each stimulus was separated by a randomly-determined interstimulus interval ranging from 1 to 10 s. At the
J.A. Caldwell et al. / Clinical Neurophysiology 114 (2003) 23–31
25
Table 1 Testing schedule Day 6 Block 1 01:45 TESTING
07:00 SLEEP
15:00 AWAKEN
05:00
09:45 EEG4 PVT4
17:45 EEG1 PVT1
Block 2 19:45 EEG1 PVT1
21:45 EEG2 PVT2
15:45 EEG5 PVT5
17:45 EEG6 PVT6
23:45 EEG2 PVT2
Day 7 Block 3 01:45 EEG3 PVT3
Block 4 03:45 EEG3 PVT3
Block 5 11:45 EEG4 PVT4
conclusion of each 10 min PVT trial, data were downloaded to a computer for analysis. 2.2.2. Electroencephalography EEG data were recorded with a Cadwell Spectrum 32 (Cadwell Laboratories, Kennewick, WA) from FZ, CZ, and PZ (referenced to linked mastoids A1, A2) throughout each 10 min PVT via Grass E5SH silver cup electrodes filled with SigmaGel electrolyte. In actuality, more than the midline electrodes were originally examined, but since there were no indications of asymmetrical effects, only the midline sites are reported for the sake of brevity. EEG data were digitized at a rate of 200 Hz and stored on optical disk for analyses. The filters were set at 0.53–70 Hz with the 60 Hz notch filter employed. Electrodes were checked before each testing session to ensure that impedances were 5000 Ohms or less. Absolute EEG power was derived for delta (1.5–3.0 Hz), theta (3.0–8.0 Hz), alpha (8.0–13.0 Hz), and beta (13.0–20.0 Hz) via fast Fourier transformations (FFT). A Hamming window was used, and the frequency bin resolution was 0.39 Hz. 2.3. Procedure 2.3.1. Selection of data for analysis As noted earlier, the data examined in this report were collected as part of a larger 10 day study; however, only the data collected on days 6 and 7 were examined here because these were the only days on which the volunteers were severely fatigued (kept awake for 32 h), and because the objective of this report was to describe the effects of postural manipulations on the EEG activity and performance of sleepdeprived people. During days 6 and 7 of the study, the volunteers were kept awake for a continuous 32 h to affect a transition from the night shift (with sleep during the day) to the day shift (with sleep at night). Specifically, the volunteers were kept awake from the end of their last day-sleep period (at 15:00 h) until 23:00 h the next night (at which time they were allowed to sleep at the more normal time from 23:00 to 07:00 h). Prior to their last day-sleep episode which began this transition, half of the volunteers received 30 mg temazepam and half received placebo (the dose was administered at 06:30 h, before the 07:00–15:00 h sleep period). The data from both
13:45 EEG5 PVT5
Block 6 19:45 EEG6 PVT6
23:00 SLEEP
drug groups were collapsed together after it was determined from a data analysis comparing the two groups that out of 48 possible drug effects on EEG activity, which could have confounded the present analysis, there was only a single significant interaction (a 3 way), and, out of 8 possible confounding drug effects on the PVT data, none were significant. 2.3.2. Testing schedule Twelve EEG/PVT test sessions were conducted. Two tests were performed in each of 6 test blocks of 4 h. The actual test times were 17:45 and 19:45 h (day 6, block 1), 21:45 and 23:45 h (day 6, block 2), 01:45 and 03:45 h (day 7, block 3), 09:45 and 11:45 h (day 7, block 4), 13:45 and 15:45 h (day 7, block 5), and 17:45 and 19:45 h (day 7, block 6) (see Table 1). In each test block, one 10 min EEG/PVT session was conducted while the volunteer was seated and one EEG/PVT session (2 h later) was conducted while the volunteer was standing. The posture order was counterbalanced between subjects (i.e. half of the subjects were tested while sitting and then while standing, and the other half were tested in the opposite order). In each of the testing sessions, participants were instructed to remain relaxed but awake throughout testing. In addition, they were instructed to hold the PVT unit with both hands, approximately 2 feet (0.6 m) from their faces. 2.4. Data analysis EEG records were visually scanned to select artifact-free epochs upon which power spectral analyses were performed. Fifteen 2.5 s epochs were selected from each of the EEG/PVT sessions. These 15 epochs were averaged together to yield one set of absolute power values for each EEG band (delta, theta, alpha, and beta) for each 10 min session. Epochs were chosen which reflected the predominant EEG activity within the majority of the 10 min period. The person choosing the epochs was blind to the postural condition and block at which the data were collected. To ensure adequate counterbalancing, the two sessions (one sitting and one standing) within each testing period were treated as a single overall block. These blocks were labeled according to the median time between the sitting and standing sessions within each block. Thus, for analysis purposes, there
26
J.A. Caldwell et al. / Clinical Neurophysiology 114 (2003) 23–31
3.1. PVT data
Table 2 Posture effects from EEG and PVT analysis PVT
Block-by-posture
Posture
a
RRT (,0.0001) a % Lapses (0.0006) RRT (0.0073) % Lapses (0.0029)
EEG Theta
Alpha
CZ (0.0238) PZ (0.0404) FZ (0.0215) CZ (0.0050) PZ (0.0083)
PZ (0.0497)
Values in parentheses are P values.
were 6 overall testing blocks (at 18:45 and 22:45 h on day 6, and at 02:45, 10:45, 18:45, and 20:45 h on day 7). Analyses were conducted to determine if differences occurred between the two sessions (with posture counterbalanced) since they were 2 h apart. However, no differences were found. The 2 h interval was selected to minimize the potential influence of boredom on task performance and/or alertness. Data were analyzed with BMDP4V, repeated measures analysis of variance (ANOVA). The factors for both PVT data (reaction times and response lapses) and EEG data (delta, theta, alpha, and beta power) were testing block (blocks 1–6) and posture (sitting versus standing). Higher-order interactions were explored via analysis of simple effects (Winer, 1962). Significant session main effects were examined for the presence of linear, cubic, and quadratic trends. Huynh–Feldt adjusted degrees of freedom were employed to correct for violations of the compound symmetry assumption.
3. Results A quick overview of the most important posture effects is presented in Table 2. Specific details on all of the collected data are presented afterwards.
The PVT data consisted of response times (from stimulus onset to the press of the response button) and the percentage of response lapses (the number of stimuli to which the response was greater than 500 ms or there was no response, divided by the total number of stimuli, multiplied by 100). Of the 16 subjects who contributed PVT data, one was excluded from analysis because his responses were 3 or more standard deviations from the group mean. Thus, the total sample size was 15. Reaction time (RT) data were converted to reciprocal reaction times (RRTs) to correct for marked deviations from normality. 3.1.1. RRT There was a block-by-posture interaction in RRT (Fð5; 70Þ ¼ 9:18, P , 0:0001) as shown in Fig. 1 (left side). Analysis of simple effects indicated an absence of differences between sitting and standing during the first part of the deprivation period (up to 12 h awake), whereas RTs were significantly slower under sitting than standing at 10:45, 14:45, and 18:45 h – the blocks corresponding to 20, 24, and 28 h awake (P , 0:05) (again, the data depicted in Fig. 1 are reciprocal transformations). This same block-byposture interaction, examined in another way, indicated that there were block differences both under the sitting condition (P , 0:05) and under the standing condition (P , 0:05). While trend analyses revealed significant linear slowing of RT as a function of testing block, or sleep deprivation, under both conditions (Fð1; 14Þ ¼ 38:52, P , 0:0001, and Fð1; 14Þ ¼ 74:05, P , 0:0001, respectively), the magnitude of this effect was greater under sitting than standing, as shown in Fig. 1. There was a block main effect (with posture collapsed) (Fð5; 70Þ ¼ 18:94, P , 0:0001) due to a significant linear trend (Fð1; 14Þ ¼ 79:42, P , 0:0001), showing a slowing of RTs from the first to the last testing block (see Fig. 2, left side). Neither the cubic nor the quadratic trends were significant. The posture main effect (with testing block
Fig. 1. The combined effects of sleep deprivation and posture on RRTs and the percentage of response lapses. The positive influence of posture became evident only after 20 h of continuous wakefulness (when alertness began to suffer). Significant differences between sitting and standing are denoted by asterisks (P , 0:05).
J.A. Caldwell et al. / Clinical Neurophysiology 114 (2003) 23–31
27
Fig. 2. The effects of sleep deprivation on RRTs and the percentage of response lapses. Linear trends were found in both, and RRT additionally displayed quadratic trends (P , 0:05).
collapsed) revealed slower overall RTs under the sitting versus the standing condition (Fð1; 14Þ ¼ 9:84, P ¼ 0:0073). The mean RRT while standing was 4.48, and the mean RRT while sitting was 4.18 (these equate to untransformed RTs of 251 and 384 ms, respectively).
3.1.2. Lapses A block-by-posture interaction occurred in the percentage of response lapses (Fð5; 70Þ ¼ 4:99, P ¼ 0:0006). Followup analyses showed that this was due to postural effects at 10:45, 14:45, and 18:45 h (P , 0:05). In these cases, the
Fig. 3. Effects of sleep deprivation on delta, theta, and alpha EEG activity recorded from midline electrodes during administration of the 10 min PVTs. Notice that absolute power increased as a function of time across all wavelengths. Data here are expressed in mV 2.
28
J.A. Caldwell et al. / Clinical Neurophysiology 114 (2003) 23–31
Fig. 4. The combined effects of sleep deprivation and posture on EEG theta activity recorded from midline electrodes during administration of the 10 min PVTs. Notice that the beneficial effects of posture became more pronounced as the degree of sleepiness increased. Data here are expressed in mV 2. Significant differences between sitting and standing are denoted by asterisks (P , 0:05).
standing condition evidenced a lower percentage of lapses, and the standard errors were reduced as well (see Fig. 1, right side). Although there were block effects both at sitting and standing (P , 0:05), the magnitude of the effect was smaller under the standing condition. Trend analyses further revealed linear increases in lapses under both postural conditions (Fð1; 14Þ ¼ 9:48, P ¼ 0:0082, and Fð1; 14Þ ¼ 17:91, P ¼ 0:0008, respectively), but this increase was attenuated under the standing condition, as shown by the presence of a smaller F value in the linear component, and the presence of a quadratic trend only under the standing intervention (Fð1; 14Þ ¼ 6:81, P ¼ 0:0206). An overall block main effect (with sitting and standing combined) was observed as well (Fð5; 70Þ ¼ 8:85, P , 0:0001). This was due to a significant linear increase in lapses throughout the sleep deprivation period (Fð1; 14Þ ¼ 21:56, P ¼ 0:0004). Means across all testing blocks are shown in Fig. 2 (right side). A posture main effect was due to better performance in the standing versus the sitting condition (Fð1; 14Þ ¼ 12:95, P ¼ 0:0029). The mean percentage of lapses was 0.67% while standing and 1.66% while sitting.
3.2. EEG 3.2.1. Delta There were no block-by-posture interactions or posture main effects within the delta band at any of the recording sites; however, there was an overall block main effect at FZ (Fð5; 75Þ ¼ 5:84, P ¼ 0:0001), CZ (Fð5; 75Þ ¼ 8:94, P , 0:0001), and PZ (Fð5; 75Þ ¼ 5:93, P ¼ 0:0001). Trend analysis indicated that these were due to a progressive linear increase in delta throughout the testing period for FZ (Fð1; 15Þ ¼ 18:31, P ¼ 0:0007), CZ (Fð1; 15Þ ¼ 28:22, P ¼ 0:0001), and PZ (Fð1; 15Þ ¼ 13:98, P ¼ 0:0020). The means across all testing blocks are shown in Fig. 3. 3.2.2. Theta Block-by-posture interactions in theta activity were observed at sites CZ (Fð5; 75Þ ¼ 2:77, P ¼ 0:0238) and PZ (Fð5; 75Þ ¼ 2:46, P ¼ 0:0404). Analysis of simple effects indicated that these were all due to a difference between sitting and standing postures at block 3 (12 h awake; CZ only), block 4 (20 h awake), and block 6 (28 h awake), with the amount of theta greater under sitting than standing.
J.A. Caldwell et al. / Clinical Neurophysiology 114 (2003) 23–31
In addition, it was evident that although there were block differences under both the sitting and standing conditions (P , 0:05), the linear increase in theta at CZ (Fð1; 15Þ ¼ 51:85, P , 0:0001) and PZ (Fð1; 15Þ ¼ 29:94, P ¼ 0:0001) while sitting was smaller in magnitude than the linear increase in theta at CZ (Fð1; 15Þ ¼ 12:18, P ¼ 0:0033) and PZ (Fð1; 15Þ ¼ 9:67, P ¼ 0:0072) while standing. Also, a subsequent attenuation of theta activity toward the end of testing (a quadratic trend noted at CZ) was only observed under the standing condition. These effects are shown in Fig. 4. Block main effects were observed at FZ (Fð5; 75Þ ¼ 18:84, P , 0:0001), CZ (Fð5; 75Þ ¼ 17:75, P , 0:0001), and PZ (Fð5; 75Þ ¼ 12:34, P , 0:0001). Trend analysis indicated that there were deprivation-related linear increases in theta at all 3 sites: FZ (Fð1; 15Þ ¼ 63:18, P , 0:0001), CZ (Fð1; 15Þ ¼ 42:79, P , 0:0001), and PZ (Fð1; 15Þ ¼ 40:72, P , 0:0001). There were also cubic trends at FZ (Fð1; 15Þ ¼ 5:38, P ¼ 0:0349) and CZ (Fð1; 15Þ ¼ 7:85, P ¼ 0:0134). As shown in Fig. 3, the cubic trends were largely due to a relatively sharp rise in theta activity between the third and fourth testing blocks (12 and 20 h awake). Posture main effects were found at all recording sites: FZ (Fð1; 15Þ ¼ 6:58, P ¼ 0:0215), CZ (Fð1; 15Þ ¼ 10:82, P ¼ 0:0050), and PZ (Fð1; 15Þ ¼ 9:23, P ¼ 0:0083). These were due to reduced theta activity while standing as opposed to sitting. 3.2.3. Alpha and beta There were no block-by-posture interactions in either the alpha or beta bands. Block main effects were observed only in the alpha range at FZ (Fð4:16; 62:33Þ ¼ 4:44, P ¼ 0:0029) and CZ (Fð3:22; 48:37Þ ¼ 3:26, P ¼ 0:0266). The effects at FZ were due to the presence of both linear (Fð1; 15Þ ¼ 5:04, P ¼ 0:0404) and quadratic trends (Fð1; 15Þ ¼ 16:30, P ¼ 0:0011), while the effect at CZ was due to a cubic trend (Fð1; 15Þ ¼ 6:29, P ¼ 0:0242). As can be seen in Fig. 3, alpha at FZ remained fairly constant across the first 3 test blocks, increased substantially from 02:45 to 10:45 h, and then leveled off for the final 3 blocks. The alpha activity at CZ behaved similarly except that alpha actually declined during the first half of the deprivation period before increasing toward the end. Posture effects occurred in the alpha band only at PZ (Fð1; 15Þ ¼ 4:55, P ¼ 0:0497), with higher mean alpha activity while standing than while sitting (means are 12.60 and 10.73, respectively). No block or posture effects occurred in the beta band. 4. Discussion This study in large part confirmed our earlier findings that body posture exerts an influence on the EEG activity of sleep-deprived volunteers (Caldwell et al., 2000a). Specifically, it was found that standing versus sitting significantly attenuated sleep deprivation-related increases in EEG theta
29
activity. In addition, the administration of the PVT, concurrent with EEG data collection, permitted confirmation of our earlier hypothesis that postural manipulations which attenuated slow-wave EEG would also mitigate sleep deprivationrelated decrements in cognitive performance (such as slower RTs and increased attentional lapses). Here it was found that standing versus sitting attenuated EEG theta activity, maintained simple RT, and sustained attention closer to well-rested levels despite 28 h of continuous wakefulness. Although changes in EEG theta activity can be interpreted as something other than a reflection of an individual’s state of alertness (for instance, a relationship has been observed between theta and cognitive workload), experimental evidence suggests that theta is in fact a reliable indicator of the gross inhibition of CNS excitation associated with low arousal and diminished information processing (Ray, 1990). In the present context, this latter interpretation (related to alertness) makes the most sense because it is more consistent with both the experimental paradigm and the behavioral evidence. With regard to the experimental paradigm, it is clear that the amount of theta became more pronounced as the period of continuous wakefulness progressed, despite the fact that the cognitive workload requirements did not change (the same task was repeated throughout). With regard to the behavioral evidence, a generalized and progressive decrement in cognitive performance likewise coincided with the increased amount of sleep deprivation, especially in the sitting condition. In addition, the upright postural manipulation that improved an EEG-based indicator of arousal likewise improved a behavioral measure of sustained attention. Based on what is known about sleep loss, physiological arousal, performance, and postural effects, both the EEG and performance findings of the present study make sense. In sleep deprivation paradigms, it has been well established that prolonged wakefulness is accompanied by systematic elevations in slow-wave delta and/or theta activity (Pigeau et al., 1987; Caldwell et al., 2002). In addition, it has been shown that sleep loss is associated with attention deficits (increased response failures or lapses), impaired information processing, slower RTs, and a host of other problems (Dinges, 1995). The fact that slower EEG activity is linked to sustained attention decrements has been empirically established as well (Belyavin and Wright, 1987). Because of this, it is reasonable that any manipulation capable of attenuating delta or theta activity also would, to some degree, mitigate cognitive decrements in sleepy volunteers. Since our previous research had shown an improvement in EEG activation when standing versus sitting (Caldwell et al., 2000a), and since Cole (1989) had earlier demonstrated similar effects from tilting volunteers upright on a tilt table, it was clear that postural changes were able to influence the EEG (and by inference, arousal) of human volunteers. Thus, it was anticipated that a more upright posture would benefit sustained attention as well, particularly in situations where
30
J.A. Caldwell et al. / Clinical Neurophysiology 114 (2003) 23–31
alertness was impaired by sleep loss. The present results show that in fact, EEG activation was maintained and attention was sustained, despite sleep loss, by requiring volunteers to transition from a sitting posture to an upright, standing posture. The effects of this postural manipulation were particularly noticeable after 20 h or more of continuous wakefulness. Consistent with our earlier findings, the postural attenuation of EEG slow-wave activity in the standing condition versus the sitting condition tended to be more noticeable as the amount of sleep loss progressed. There were no posturally-related differences in theta activity after only 4 and 8 h of continuous wakefulness, and only one difference (in CZ theta) after 12 h of wakefulness. However, there was significantly less theta recorded from all 3 electrodes (FZ, CZ, and PZ) under the sitting versus the standing condition at the 20 and 28 h points. It is interesting to note that a somewhat similar effect has been observed in psychostimulant research where even amphetamine fails to reliably enhance performance, mood, and EEG activity until a substantial amount of sleepiness is present (Caldwell et al., 2000b). Apparently there is a ‘ceiling effect’ which makes it difficult to demonstrate enhanced alertness/performance in people who are already fully alert. However, once sleepiness is present, the benefits of effective alertness-promoting strategies are more obvious. The data from this study show that an upright postural orientation is one such strategy in terms of sustaining both the EEG activation and sustained attention of sleepy individuals. This may be part of the reason that the military has long used postural interventions in situations requiring vigilance, as evidenced by the term ‘standing watch’. One difference between the present results and our earlier findings (Caldwell et al., 2000a) was that we originally found postural effects in both the delta and theta EEG bands, whereas the present study revealed effects only in the theta band. This could have been due to the fact that our previous investigation required participants to remain passive during testing, whereas this study required the participants to actively engage in a short task (the PVT). Since contextual stimulation (cognitive, sensory, etc.) is known to attenuate the effects of sleep deprivation (Dijkman et al., 1997), and since our present subjects were presumably motivated to complete the task at hand, the impact of sleepiness on alertness may have been reduced, thus preventing frank sleep episodes (delta activity) while still not altogether eliminating slow-wave EEG activity. In conclusion, it should be noted that while an upright postural orientation appears useful for improving alertness in sleep-deprived volunteers, the duration of this effect is unknown. In the present study, test sessions lasted only 10 min, so it is possible that the effects could be short-lived. A future study should investigate the time course of postural effects using 20 or 30 min tasks. In addition, it should be noted that this study could not resolve whether the beneficial effects of standing up were associated with posturally-
induced hemodynamic changes (as postulated by Cole, 1989), core temperature changes (as reported by Kra¨ uchi et al., 1997), or some other mechanism since blood pressure, temperature, or other physiological measures (other than EEG) were not collected. However, regardless of the physiological basis for the observed effects, it seems clear that standing upright attenuates the impact of sleepiness. For this reason, postural countermeasures should be recommended along with more traditional strategies (such as caffeine ingestion, activity breaks, etc.) for promoting the alertness and performance of sleepy people.
References Aeschbach D, Cajochen C, Tobler I, Dijk D, Borbe´ ly AA. Sleep in a sitting position: effect of triazolam on sleep stages and EEG power spectra. Psychopharmacology 1994;14:209–214. Belyavin A, Wright NA. Changes in electrical activity of the brain with vigilance. Electroenceph clin Neurophysiol 1987;66:137–144. Bonnet MH. Sleep deprivation. In: Kryger MH, Roth T, Dement WC, editors. Principles and practice of sleep medicine, 3rd ed. Philadelphia, PA: W.B. Saunders, 2000. pp. 53–71. Caldwell JA, Prazinko BF, Hall KK. The effects of body posture on resting electroencephalographic activity in sleep-deprived subjects. Clin Neurophysiol 2000a;111:464–470. Caldwell JA, Smythe NK, LeDuc PA, Caldwell JL. Efficacy of Dexedrine w for maintaining aviator performance during 64 hours of sustained wakefulness. Aviat Space Environ Med 2000b;71(1):7–18. Caldwell JL, Hall KK, Prazinko BF, Norman DA, Rowe TL, Erickson BS, Estrada A, Caldwell JA. The efficacy of temazepam for improving daytime sleep and nighttime performance in army aviators, USAARL Report, No. 2002-05, Fort Rucker, AL: US Army Aeromedical Research Laboratory, 2001. Caldwell JA, Hall KK, Erickson BS. EEG data collected from helicopter pilots in flight are sufficiently sensitive to detect increased fatigue from sleep deprivation. Int J Aviat Psychol 2002;12(1):19–32. Cole RJ. Postural baroreflex stimuli may affect EEG arousal and sleep in humans. J Appl Phys 1989;67(6):2369–2375. Dijkman M, Sachs N, Levine E, Mallis M, Carlin MM, Gillen KA, Powell JW, Samuel S, Mullington J, Rosekind MR, Dinges D. Effects of reduced stimulation on neurobehavioral alertness depend on circadian phase during human sleep deprivation. Sleep Res 1997;26:265. Dinges DF. An overview of sleepiness and accidents. J Sleep Res 1995;4(Suppl 2):4–14. Dinges DF, Pack F, Williams K, Gillen KA, Powell JW, Ott GE, Aptowicz C, Pack AI. Cumulative sleepiness, mood disturbance, and psychomotor vigilance performance decrements during a week of sleep restricted to 4-5 hours per night. Sleep 1997;20(4):267–277. Goldstein L, Murphree HB, Pfeiffer CC. Comparative study of EEG effects of antihistamines in normal volunteers. J Clin Pharm 1968;Jan/Feb:42– 53. Hickler RB, Wells RE, Tyler HR, Hamlin JT. Plasma catecholamine and electroencephalographic responses to acute postural change. Am J Med 1959;26:410–422. Kleitman N, Doktorsky A. Studies on the physiology of sleep VII. The effect of the position of the body and of sleep on rectal temperature in man. Am J Physiol 1933;104:340–343. Kra¨ uchi K, Cajochen C, Wirz-Justice A. A relationship between heat loss and sleepiness: effects of postural change and melatonin administration. J Appl Physiol 1997;83(1):134–139. Nicholson AN, Stone BM. Influence of back angle on the quality of sleep in seats. Ergonomics 1987;30(7):1033–1041. Pigeau RA, Heslegrave RJ, Angus RG. Psychophysiological measures of
J.A. Caldwell et al. / Clinical Neurophysiology 114 (2003) 23–31 drowsiness as estimators of mental fatigue and performance degradation during sleep deprivation. Electric and magnetic activity of the central nervous system: research and clinical applications in aerospace medicine, AGARD CP-432, Neuilly sur Seine: Advisory Group for Aerospace Research and Development, 1987. pp. 21-1–21-16. Ray WJ. The electrocortical system. In: Cacioppo JT, Tassinary LG, editors. Principles of psychophysiology, New York: Cambridge University Press, 1990. pp. 385–412. Sterman MB, Schummer GJ, Dushenko TW, Smith JC. Electroencephalographic correlates of pilot performance: simulation and in-flight studies. Electrical and magnetic activity of the central nervous system: research and clinical applications in aerospace medicine, AGARD CP-432,
31
Neuilly sur Seine: North Atlantic Treaty Organization, 1987. pp. 311–31-16. Sundin T. The influence of body position on the urinary excretion of adrenaline and noradrenaline. Acta Med Scand (Suppl) 1956;313:1–10. Tikuisis P, Ducharme MB. The effect of postural changes on body temperatures and heat balance. Eur J Appl Physiol 1996;72:451–455. Wilson GF, Hankins T. EEG and subjective measures of private pilot workload. Proceedings of the Human Factors and Ergonomics Society 38th annual meetingSanta Monica, CA: Human Factors and Ergonomics Society, 1994. pp. 1322–1325. Winer BJ. Statistical principles in experimental design, New York: McGraw-Hill, 1962.
Body posture affects electroencephalographic activity and psychomotor vigilance task performance in sleep-deprived subjects John A. Caldwell* ,1, Brian Prazinko, J. Lynn Caldwell 1 Aeromedical Research Laboratory, Fort Rucker, AL 36362-0577, USA Accepted 12 August 2002
Abstract Objective: This study examined the effects of posture on electroencephalographic (EEG) activity and psychomotor vigilance task (PVT) performance in 16 sleep-deprived volunteers. Methods: EEG data were collected while participants completed 10 min PVTs under two counterbalanced sitting/standing conditions during 28 h of continuous wakefulness. Results: In both the sitting and standing conditions, theta activity progressively increased as a function of sleep loss, but standing upright significantly attenuated this effect, suggesting that alertness was improved by the more upright posture. The PVT results showed that cognitive psychomotor performance was maintained at nearly well-rested levels by standing upright, whereas reaction time and attention noticeably deteriorated when participants were seated. Conclusions: These results suggest that an upright posture increases EEG arousal and sustained attention, indicating that postural manipulations can be useful for counteracting fatigue in sleep-deprived individuals. q 2002 Published by Elsevier Science Ireland Ltd. Keywords: Body posture; Electroencephalographic; Sleep deprivation; Psychomotor vigilance task; Performance
1. Introduction Although there is little direct evidence that posture influences the alertness of individuals who are fatigued, the scientific literature and anecdotal reports suggest it is easier to maintain wakefulness when sitting or standing than when lying down (Bonnet, 2000). Shift workers, who suffer from drowsiness on the job due to chronically shortened or disrupted sleep (Dinges, 1995), often state that standing up is one way of improving alertness at work, and it has been found that standing as opposed to lying down contributes to enhanced alertness, possibly by altering afferent sensory input (Dijkman et al., 1997). In fact, several studies have suggested that a more upright postural orientation actually inhibits sleepiness. Nicholson and Stone (1987) found that subjects experienced reductions in total sleep time, decreased sleep efficiency, and increased awakenings when they attempted to sleep in a more-upright sitting position (17.58 from the vertical angle), as opposed to either lying flat or reclining 49.58 or 378. Similar results * Corresponding author. Tel.: 11-210-536-8134. E-mail address: [email protected] (J.A. Caldwell). 1 Present address: U.S. Air Force Research Laboratory, 2485 Gillingham Drive, Suite 25, Brooks AFB, TX 78235, USA.
were reported by Aeschbach et al. (1994) who discovered that subjects who slept in reclining chairs (rather than lying flat in bed) experienced reduced sleep efficiency, less rapideye-movement (REM) sleep, and increased stage 1 sleep. The reasons for these postural effects are not completely understood, but changes in core body temperature, plasma catecholamine levels, and hemodynamic mechanisms may be involved. With regard to the role of body temperature, Kleitman and Doktorsky (1933) long ago reported that subjects manifested higher rectal temperatures when standing upright than when supine and vice versa, and these findings have been confirmed by Tikuisis and Ducharme (1996) and Kra¨uchi et al. (1997). Furthermore, changes in core body temperature have been associated with alterations in self-reported sleepiness – Kra¨uchi et al. (1997) found that higher core body temperatures were accompanied by lower self-ratings of tiredness (sleepiness) in subjects who were sitting/standing rather than lying down. With regard to the role of hemodynamic and/or catecholaminergic mediators, it is well known that standing up produces elevations in both heart rate and blood pressure which counteract the loss in cerebral blood flow stemming from an initial drop in venous pressure. Levels of urinary epinephrine and norepinephrine have been found to rise as a result of these hemodynamic changes (Sundin, 1956), and augmented adrenergic nerve
1388-2457/02/$ - see front matter q 2002 Published by Elsevier Science Ireland Ltd. PII: S 1388-245 7(02)00283-3
CLINPH 2002093
24
J.A. Caldwell et al. / Clinical Neurophysiology 114 (2003) 23–31
activity and stimulation of the adrenal medulla follows. Hickler et al. (1959) showed that homeostatically-driven elevations in heart rate are accompanied by a significant rise in the levels of plasma catecholamines. In addition to posturally-induced peripheral and neurendocrine effects, hemodynamic changes have been shown to influence central nervous system (CNS) activation. Cole (1989) found that heart rate and blood pressure elevations were accompanied by an increase in the amplitude-percentage product of high-frequency (beta) electroencephalographic (EEG) activity when subjects were tilted 408 upright on a tilt table. The upright tilt also inhibited sleepiness as evidenced by the fact that volunteers were able to stave off sleep for almost 6 min longer than when they were tested in the supine position. It was concluded that upright posture enhanced arousal via hemodynamically-induced reductions in baroreceptor firing (subjects were tested both with and without the application of positive leg pressure in an effort to isolate the effects of the baroreflex response). These findings suggest the importance of postural effects on sleepiness – a relationship which has been recognized by circadian rhythm/sleep deprivation researchers who have implemented ‘constant routine’ protocols in which subjects are prevented from standing and moving about while undergoing laboratory testing. However, the isolated impact of posture has not been well-studied in sleep-deprived individuals. Furthermore, the feasibility of using administrativelyimposed or self-imposed postural changes as a countermeasure for drowsiness in sleepy shift workers or other sleepdeprived personnel has not been established. In a previous study (Caldwell et al., 2000a), we found that requiring sleep-deprived volunteers to stand up, rather than remain seated, attenuated the fatigue-related elevations in both delta (1.5–3.0 Hz) and theta (3.0–8.0 Hz) activity that are commonly observed in sleepy individuals. We further found that this effect persisted throughout more than 55 h of continuous wakefulness. Although increased theta activity has been found to coincide with increased workload (Sterman et al., 1987; Wilson and Hankins, 1994) as well as increased fatigue, the widespread augmentation of theta activity has more traditionally been associated with impaired alertness and drug-induced sedation (Goldstein et al., 1968; Pigeau et al., 1987), especially in contexts where either of these factors were known to be present. In addition, these generalized low-frequency EEG patterns have been associated with decrements on vigilance tasks (Belyavin and Wright, 1987). Since this has been the case, we hypothesized that the apparent CNS stimulation offered by an upright postural intervention (Caldwell et al., 2000a) should also benefit cognitive psychomotor performance. However, no performance measures were collected in the earlier study. Because of this, the present investigation was performed in an effort to explore the effects of posture (sitting versus standing) both on EEG activity and concurrently-measured psychomotor vigilance task (PVT) performance. Specifically, it was of interest to learn whether the
apparent improvement in physiological arousal associated with standing (versus sitting) would translate into improved sustained attention. 2. Methods The data reported here were selected from a larger study in which the efficacy of temazepam was assessed for promoting daytime sleep in shift workers (further information is provided below). The entire study is explained in Caldwell et al. (2001). This research was approved by the U.S. Army Aeromedical Research Laboratory’s (USAARL’s) local Scientific and Human Use Committees prior to execution. In addition, the protocol was approved by the Human Subjects Review Board of the U.S. Army Medical Research and Materiel Command. 2.1. Subjects Fifteen males and one female resided at USAARL for a total of 10 days (although only 2 of these days are the focus of this report). The mean age of the volunteers was 33.7 years, with a range of 26–44 years. All subjects were medically evaluated prior to the study, and none were found to be experiencing any current significant illnesses. None of the participants had any evidence of past psychiatric problems or sleep disorders. Volunteers were not allowed to consume alcohol, caffeine, or any type of medication (other than acetaminophen and ibuprofen) during the protocol. Caffeine users were asked to eliminate caffeine consumption several days prior to the study. Only one subject used nicotine; he was allowed to smoke between test sessions, but not during testing. Given the time of the EEG sessions, the last cigarette before testing would have been no earlier than 45 min prior to the test. There were 3 left-handed subjects, and the remainder were right-handed. Each subject signed an informed consent agreement prior to their participation. 2.2. Apparatus 2.2.1. PVT The PVT (Ambulatory Monitoring, Inc., Ardsley, NY) is a portable, simple reaction time test known to be sensitive to sleep loss (Dinges et al., 1997). A 10 min trial of the PVT was used at each testing time in this investigation. The device visually displays numbers (milliseconds) that increment rapidly in sequence for up to 1.5 s until the volunteer responds by pressing a microswitch. The number of milliseconds from stimulus onset to response was recorded. At the outset, participants were admonished to respond to the stimulus as quickly as possible. If the participant responded later than 500 ms or failed to respond within 1.5 s, a ‘lapse’ was recorded. Since the time of this task was set for 10 min, the actual number of stimuli presented varied within each trial since each stimulus was separated by a randomly-determined interstimulus interval ranging from 1 to 10 s. At the
J.A. Caldwell et al. / Clinical Neurophysiology 114 (2003) 23–31
25
Table 1 Testing schedule Day 6 Block 1 01:45 TESTING
07:00 SLEEP
15:00 AWAKEN
05:00
09:45 EEG4 PVT4
17:45 EEG1 PVT1
Block 2 19:45 EEG1 PVT1
21:45 EEG2 PVT2
15:45 EEG5 PVT5
17:45 EEG6 PVT6
23:45 EEG2 PVT2
Day 7 Block 3 01:45 EEG3 PVT3
Block 4 03:45 EEG3 PVT3
Block 5 11:45 EEG4 PVT4
conclusion of each 10 min PVT trial, data were downloaded to a computer for analysis. 2.2.2. Electroencephalography EEG data were recorded with a Cadwell Spectrum 32 (Cadwell Laboratories, Kennewick, WA) from FZ, CZ, and PZ (referenced to linked mastoids A1, A2) throughout each 10 min PVT via Grass E5SH silver cup electrodes filled with SigmaGel electrolyte. In actuality, more than the midline electrodes were originally examined, but since there were no indications of asymmetrical effects, only the midline sites are reported for the sake of brevity. EEG data were digitized at a rate of 200 Hz and stored on optical disk for analyses. The filters were set at 0.53–70 Hz with the 60 Hz notch filter employed. Electrodes were checked before each testing session to ensure that impedances were 5000 Ohms or less. Absolute EEG power was derived for delta (1.5–3.0 Hz), theta (3.0–8.0 Hz), alpha (8.0–13.0 Hz), and beta (13.0–20.0 Hz) via fast Fourier transformations (FFT). A Hamming window was used, and the frequency bin resolution was 0.39 Hz. 2.3. Procedure 2.3.1. Selection of data for analysis As noted earlier, the data examined in this report were collected as part of a larger 10 day study; however, only the data collected on days 6 and 7 were examined here because these were the only days on which the volunteers were severely fatigued (kept awake for 32 h), and because the objective of this report was to describe the effects of postural manipulations on the EEG activity and performance of sleepdeprived people. During days 6 and 7 of the study, the volunteers were kept awake for a continuous 32 h to affect a transition from the night shift (with sleep during the day) to the day shift (with sleep at night). Specifically, the volunteers were kept awake from the end of their last day-sleep period (at 15:00 h) until 23:00 h the next night (at which time they were allowed to sleep at the more normal time from 23:00 to 07:00 h). Prior to their last day-sleep episode which began this transition, half of the volunteers received 30 mg temazepam and half received placebo (the dose was administered at 06:30 h, before the 07:00–15:00 h sleep period). The data from both
13:45 EEG5 PVT5
Block 6 19:45 EEG6 PVT6
23:00 SLEEP
drug groups were collapsed together after it was determined from a data analysis comparing the two groups that out of 48 possible drug effects on EEG activity, which could have confounded the present analysis, there was only a single significant interaction (a 3 way), and, out of 8 possible confounding drug effects on the PVT data, none were significant. 2.3.2. Testing schedule Twelve EEG/PVT test sessions were conducted. Two tests were performed in each of 6 test blocks of 4 h. The actual test times were 17:45 and 19:45 h (day 6, block 1), 21:45 and 23:45 h (day 6, block 2), 01:45 and 03:45 h (day 7, block 3), 09:45 and 11:45 h (day 7, block 4), 13:45 and 15:45 h (day 7, block 5), and 17:45 and 19:45 h (day 7, block 6) (see Table 1). In each test block, one 10 min EEG/PVT session was conducted while the volunteer was seated and one EEG/PVT session (2 h later) was conducted while the volunteer was standing. The posture order was counterbalanced between subjects (i.e. half of the subjects were tested while sitting and then while standing, and the other half were tested in the opposite order). In each of the testing sessions, participants were instructed to remain relaxed but awake throughout testing. In addition, they were instructed to hold the PVT unit with both hands, approximately 2 feet (0.6 m) from their faces. 2.4. Data analysis EEG records were visually scanned to select artifact-free epochs upon which power spectral analyses were performed. Fifteen 2.5 s epochs were selected from each of the EEG/PVT sessions. These 15 epochs were averaged together to yield one set of absolute power values for each EEG band (delta, theta, alpha, and beta) for each 10 min session. Epochs were chosen which reflected the predominant EEG activity within the majority of the 10 min period. The person choosing the epochs was blind to the postural condition and block at which the data were collected. To ensure adequate counterbalancing, the two sessions (one sitting and one standing) within each testing period were treated as a single overall block. These blocks were labeled according to the median time between the sitting and standing sessions within each block. Thus, for analysis purposes, there
26
J.A. Caldwell et al. / Clinical Neurophysiology 114 (2003) 23–31
3.1. PVT data
Table 2 Posture effects from EEG and PVT analysis PVT
Block-by-posture
Posture
a
RRT (,0.0001) a % Lapses (0.0006) RRT (0.0073) % Lapses (0.0029)
EEG Theta
Alpha
CZ (0.0238) PZ (0.0404) FZ (0.0215) CZ (0.0050) PZ (0.0083)
PZ (0.0497)
Values in parentheses are P values.
were 6 overall testing blocks (at 18:45 and 22:45 h on day 6, and at 02:45, 10:45, 18:45, and 20:45 h on day 7). Analyses were conducted to determine if differences occurred between the two sessions (with posture counterbalanced) since they were 2 h apart. However, no differences were found. The 2 h interval was selected to minimize the potential influence of boredom on task performance and/or alertness. Data were analyzed with BMDP4V, repeated measures analysis of variance (ANOVA). The factors for both PVT data (reaction times and response lapses) and EEG data (delta, theta, alpha, and beta power) were testing block (blocks 1–6) and posture (sitting versus standing). Higher-order interactions were explored via analysis of simple effects (Winer, 1962). Significant session main effects were examined for the presence of linear, cubic, and quadratic trends. Huynh–Feldt adjusted degrees of freedom were employed to correct for violations of the compound symmetry assumption.
3. Results A quick overview of the most important posture effects is presented in Table 2. Specific details on all of the collected data are presented afterwards.
The PVT data consisted of response times (from stimulus onset to the press of the response button) and the percentage of response lapses (the number of stimuli to which the response was greater than 500 ms or there was no response, divided by the total number of stimuli, multiplied by 100). Of the 16 subjects who contributed PVT data, one was excluded from analysis because his responses were 3 or more standard deviations from the group mean. Thus, the total sample size was 15. Reaction time (RT) data were converted to reciprocal reaction times (RRTs) to correct for marked deviations from normality. 3.1.1. RRT There was a block-by-posture interaction in RRT (Fð5; 70Þ ¼ 9:18, P , 0:0001) as shown in Fig. 1 (left side). Analysis of simple effects indicated an absence of differences between sitting and standing during the first part of the deprivation period (up to 12 h awake), whereas RTs were significantly slower under sitting than standing at 10:45, 14:45, and 18:45 h – the blocks corresponding to 20, 24, and 28 h awake (P , 0:05) (again, the data depicted in Fig. 1 are reciprocal transformations). This same block-byposture interaction, examined in another way, indicated that there were block differences both under the sitting condition (P , 0:05) and under the standing condition (P , 0:05). While trend analyses revealed significant linear slowing of RT as a function of testing block, or sleep deprivation, under both conditions (Fð1; 14Þ ¼ 38:52, P , 0:0001, and Fð1; 14Þ ¼ 74:05, P , 0:0001, respectively), the magnitude of this effect was greater under sitting than standing, as shown in Fig. 1. There was a block main effect (with posture collapsed) (Fð5; 70Þ ¼ 18:94, P , 0:0001) due to a significant linear trend (Fð1; 14Þ ¼ 79:42, P , 0:0001), showing a slowing of RTs from the first to the last testing block (see Fig. 2, left side). Neither the cubic nor the quadratic trends were significant. The posture main effect (with testing block
Fig. 1. The combined effects of sleep deprivation and posture on RRTs and the percentage of response lapses. The positive influence of posture became evident only after 20 h of continuous wakefulness (when alertness began to suffer). Significant differences between sitting and standing are denoted by asterisks (P , 0:05).
J.A. Caldwell et al. / Clinical Neurophysiology 114 (2003) 23–31
27
Fig. 2. The effects of sleep deprivation on RRTs and the percentage of response lapses. Linear trends were found in both, and RRT additionally displayed quadratic trends (P , 0:05).
collapsed) revealed slower overall RTs under the sitting versus the standing condition (Fð1; 14Þ ¼ 9:84, P ¼ 0:0073). The mean RRT while standing was 4.48, and the mean RRT while sitting was 4.18 (these equate to untransformed RTs of 251 and 384 ms, respectively).
3.1.2. Lapses A block-by-posture interaction occurred in the percentage of response lapses (Fð5; 70Þ ¼ 4:99, P ¼ 0:0006). Followup analyses showed that this was due to postural effects at 10:45, 14:45, and 18:45 h (P , 0:05). In these cases, the
Fig. 3. Effects of sleep deprivation on delta, theta, and alpha EEG activity recorded from midline electrodes during administration of the 10 min PVTs. Notice that absolute power increased as a function of time across all wavelengths. Data here are expressed in mV 2.
28
J.A. Caldwell et al. / Clinical Neurophysiology 114 (2003) 23–31
Fig. 4. The combined effects of sleep deprivation and posture on EEG theta activity recorded from midline electrodes during administration of the 10 min PVTs. Notice that the beneficial effects of posture became more pronounced as the degree of sleepiness increased. Data here are expressed in mV 2. Significant differences between sitting and standing are denoted by asterisks (P , 0:05).
standing condition evidenced a lower percentage of lapses, and the standard errors were reduced as well (see Fig. 1, right side). Although there were block effects both at sitting and standing (P , 0:05), the magnitude of the effect was smaller under the standing condition. Trend analyses further revealed linear increases in lapses under both postural conditions (Fð1; 14Þ ¼ 9:48, P ¼ 0:0082, and Fð1; 14Þ ¼ 17:91, P ¼ 0:0008, respectively), but this increase was attenuated under the standing condition, as shown by the presence of a smaller F value in the linear component, and the presence of a quadratic trend only under the standing intervention (Fð1; 14Þ ¼ 6:81, P ¼ 0:0206). An overall block main effect (with sitting and standing combined) was observed as well (Fð5; 70Þ ¼ 8:85, P , 0:0001). This was due to a significant linear increase in lapses throughout the sleep deprivation period (Fð1; 14Þ ¼ 21:56, P ¼ 0:0004). Means across all testing blocks are shown in Fig. 2 (right side). A posture main effect was due to better performance in the standing versus the sitting condition (Fð1; 14Þ ¼ 12:95, P ¼ 0:0029). The mean percentage of lapses was 0.67% while standing and 1.66% while sitting.
3.2. EEG 3.2.1. Delta There were no block-by-posture interactions or posture main effects within the delta band at any of the recording sites; however, there was an overall block main effect at FZ (Fð5; 75Þ ¼ 5:84, P ¼ 0:0001), CZ (Fð5; 75Þ ¼ 8:94, P , 0:0001), and PZ (Fð5; 75Þ ¼ 5:93, P ¼ 0:0001). Trend analysis indicated that these were due to a progressive linear increase in delta throughout the testing period for FZ (Fð1; 15Þ ¼ 18:31, P ¼ 0:0007), CZ (Fð1; 15Þ ¼ 28:22, P ¼ 0:0001), and PZ (Fð1; 15Þ ¼ 13:98, P ¼ 0:0020). The means across all testing blocks are shown in Fig. 3. 3.2.2. Theta Block-by-posture interactions in theta activity were observed at sites CZ (Fð5; 75Þ ¼ 2:77, P ¼ 0:0238) and PZ (Fð5; 75Þ ¼ 2:46, P ¼ 0:0404). Analysis of simple effects indicated that these were all due to a difference between sitting and standing postures at block 3 (12 h awake; CZ only), block 4 (20 h awake), and block 6 (28 h awake), with the amount of theta greater under sitting than standing.
J.A. Caldwell et al. / Clinical Neurophysiology 114 (2003) 23–31
In addition, it was evident that although there were block differences under both the sitting and standing conditions (P , 0:05), the linear increase in theta at CZ (Fð1; 15Þ ¼ 51:85, P , 0:0001) and PZ (Fð1; 15Þ ¼ 29:94, P ¼ 0:0001) while sitting was smaller in magnitude than the linear increase in theta at CZ (Fð1; 15Þ ¼ 12:18, P ¼ 0:0033) and PZ (Fð1; 15Þ ¼ 9:67, P ¼ 0:0072) while standing. Also, a subsequent attenuation of theta activity toward the end of testing (a quadratic trend noted at CZ) was only observed under the standing condition. These effects are shown in Fig. 4. Block main effects were observed at FZ (Fð5; 75Þ ¼ 18:84, P , 0:0001), CZ (Fð5; 75Þ ¼ 17:75, P , 0:0001), and PZ (Fð5; 75Þ ¼ 12:34, P , 0:0001). Trend analysis indicated that there were deprivation-related linear increases in theta at all 3 sites: FZ (Fð1; 15Þ ¼ 63:18, P , 0:0001), CZ (Fð1; 15Þ ¼ 42:79, P , 0:0001), and PZ (Fð1; 15Þ ¼ 40:72, P , 0:0001). There were also cubic trends at FZ (Fð1; 15Þ ¼ 5:38, P ¼ 0:0349) and CZ (Fð1; 15Þ ¼ 7:85, P ¼ 0:0134). As shown in Fig. 3, the cubic trends were largely due to a relatively sharp rise in theta activity between the third and fourth testing blocks (12 and 20 h awake). Posture main effects were found at all recording sites: FZ (Fð1; 15Þ ¼ 6:58, P ¼ 0:0215), CZ (Fð1; 15Þ ¼ 10:82, P ¼ 0:0050), and PZ (Fð1; 15Þ ¼ 9:23, P ¼ 0:0083). These were due to reduced theta activity while standing as opposed to sitting. 3.2.3. Alpha and beta There were no block-by-posture interactions in either the alpha or beta bands. Block main effects were observed only in the alpha range at FZ (Fð4:16; 62:33Þ ¼ 4:44, P ¼ 0:0029) and CZ (Fð3:22; 48:37Þ ¼ 3:26, P ¼ 0:0266). The effects at FZ were due to the presence of both linear (Fð1; 15Þ ¼ 5:04, P ¼ 0:0404) and quadratic trends (Fð1; 15Þ ¼ 16:30, P ¼ 0:0011), while the effect at CZ was due to a cubic trend (Fð1; 15Þ ¼ 6:29, P ¼ 0:0242). As can be seen in Fig. 3, alpha at FZ remained fairly constant across the first 3 test blocks, increased substantially from 02:45 to 10:45 h, and then leveled off for the final 3 blocks. The alpha activity at CZ behaved similarly except that alpha actually declined during the first half of the deprivation period before increasing toward the end. Posture effects occurred in the alpha band only at PZ (Fð1; 15Þ ¼ 4:55, P ¼ 0:0497), with higher mean alpha activity while standing than while sitting (means are 12.60 and 10.73, respectively). No block or posture effects occurred in the beta band. 4. Discussion This study in large part confirmed our earlier findings that body posture exerts an influence on the EEG activity of sleep-deprived volunteers (Caldwell et al., 2000a). Specifically, it was found that standing versus sitting significantly attenuated sleep deprivation-related increases in EEG theta
29
activity. In addition, the administration of the PVT, concurrent with EEG data collection, permitted confirmation of our earlier hypothesis that postural manipulations which attenuated slow-wave EEG would also mitigate sleep deprivationrelated decrements in cognitive performance (such as slower RTs and increased attentional lapses). Here it was found that standing versus sitting attenuated EEG theta activity, maintained simple RT, and sustained attention closer to well-rested levels despite 28 h of continuous wakefulness. Although changes in EEG theta activity can be interpreted as something other than a reflection of an individual’s state of alertness (for instance, a relationship has been observed between theta and cognitive workload), experimental evidence suggests that theta is in fact a reliable indicator of the gross inhibition of CNS excitation associated with low arousal and diminished information processing (Ray, 1990). In the present context, this latter interpretation (related to alertness) makes the most sense because it is more consistent with both the experimental paradigm and the behavioral evidence. With regard to the experimental paradigm, it is clear that the amount of theta became more pronounced as the period of continuous wakefulness progressed, despite the fact that the cognitive workload requirements did not change (the same task was repeated throughout). With regard to the behavioral evidence, a generalized and progressive decrement in cognitive performance likewise coincided with the increased amount of sleep deprivation, especially in the sitting condition. In addition, the upright postural manipulation that improved an EEG-based indicator of arousal likewise improved a behavioral measure of sustained attention. Based on what is known about sleep loss, physiological arousal, performance, and postural effects, both the EEG and performance findings of the present study make sense. In sleep deprivation paradigms, it has been well established that prolonged wakefulness is accompanied by systematic elevations in slow-wave delta and/or theta activity (Pigeau et al., 1987; Caldwell et al., 2002). In addition, it has been shown that sleep loss is associated with attention deficits (increased response failures or lapses), impaired information processing, slower RTs, and a host of other problems (Dinges, 1995). The fact that slower EEG activity is linked to sustained attention decrements has been empirically established as well (Belyavin and Wright, 1987). Because of this, it is reasonable that any manipulation capable of attenuating delta or theta activity also would, to some degree, mitigate cognitive decrements in sleepy volunteers. Since our previous research had shown an improvement in EEG activation when standing versus sitting (Caldwell et al., 2000a), and since Cole (1989) had earlier demonstrated similar effects from tilting volunteers upright on a tilt table, it was clear that postural changes were able to influence the EEG (and by inference, arousal) of human volunteers. Thus, it was anticipated that a more upright posture would benefit sustained attention as well, particularly in situations where
30
J.A. Caldwell et al. / Clinical Neurophysiology 114 (2003) 23–31
alertness was impaired by sleep loss. The present results show that in fact, EEG activation was maintained and attention was sustained, despite sleep loss, by requiring volunteers to transition from a sitting posture to an upright, standing posture. The effects of this postural manipulation were particularly noticeable after 20 h or more of continuous wakefulness. Consistent with our earlier findings, the postural attenuation of EEG slow-wave activity in the standing condition versus the sitting condition tended to be more noticeable as the amount of sleep loss progressed. There were no posturally-related differences in theta activity after only 4 and 8 h of continuous wakefulness, and only one difference (in CZ theta) after 12 h of wakefulness. However, there was significantly less theta recorded from all 3 electrodes (FZ, CZ, and PZ) under the sitting versus the standing condition at the 20 and 28 h points. It is interesting to note that a somewhat similar effect has been observed in psychostimulant research where even amphetamine fails to reliably enhance performance, mood, and EEG activity until a substantial amount of sleepiness is present (Caldwell et al., 2000b). Apparently there is a ‘ceiling effect’ which makes it difficult to demonstrate enhanced alertness/performance in people who are already fully alert. However, once sleepiness is present, the benefits of effective alertness-promoting strategies are more obvious. The data from this study show that an upright postural orientation is one such strategy in terms of sustaining both the EEG activation and sustained attention of sleepy individuals. This may be part of the reason that the military has long used postural interventions in situations requiring vigilance, as evidenced by the term ‘standing watch’. One difference between the present results and our earlier findings (Caldwell et al., 2000a) was that we originally found postural effects in both the delta and theta EEG bands, whereas the present study revealed effects only in the theta band. This could have been due to the fact that our previous investigation required participants to remain passive during testing, whereas this study required the participants to actively engage in a short task (the PVT). Since contextual stimulation (cognitive, sensory, etc.) is known to attenuate the effects of sleep deprivation (Dijkman et al., 1997), and since our present subjects were presumably motivated to complete the task at hand, the impact of sleepiness on alertness may have been reduced, thus preventing frank sleep episodes (delta activity) while still not altogether eliminating slow-wave EEG activity. In conclusion, it should be noted that while an upright postural orientation appears useful for improving alertness in sleep-deprived volunteers, the duration of this effect is unknown. In the present study, test sessions lasted only 10 min, so it is possible that the effects could be short-lived. A future study should investigate the time course of postural effects using 20 or 30 min tasks. In addition, it should be noted that this study could not resolve whether the beneficial effects of standing up were associated with posturally-
induced hemodynamic changes (as postulated by Cole, 1989), core temperature changes (as reported by Kra¨ uchi et al., 1997), or some other mechanism since blood pressure, temperature, or other physiological measures (other than EEG) were not collected. However, regardless of the physiological basis for the observed effects, it seems clear that standing upright attenuates the impact of sleepiness. For this reason, postural countermeasures should be recommended along with more traditional strategies (such as caffeine ingestion, activity breaks, etc.) for promoting the alertness and performance of sleepy people.
References Aeschbach D, Cajochen C, Tobler I, Dijk D, Borbe´ ly AA. Sleep in a sitting position: effect of triazolam on sleep stages and EEG power spectra. Psychopharmacology 1994;14:209–214. Belyavin A, Wright NA. Changes in electrical activity of the brain with vigilance. Electroenceph clin Neurophysiol 1987;66:137–144. Bonnet MH. Sleep deprivation. In: Kryger MH, Roth T, Dement WC, editors. Principles and practice of sleep medicine, 3rd ed. Philadelphia, PA: W.B. Saunders, 2000. pp. 53–71. Caldwell JA, Prazinko BF, Hall KK. The effects of body posture on resting electroencephalographic activity in sleep-deprived subjects. Clin Neurophysiol 2000a;111:464–470. Caldwell JA, Smythe NK, LeDuc PA, Caldwell JL. Efficacy of Dexedrine w for maintaining aviator performance during 64 hours of sustained wakefulness. Aviat Space Environ Med 2000b;71(1):7–18. Caldwell JL, Hall KK, Prazinko BF, Norman DA, Rowe TL, Erickson BS, Estrada A, Caldwell JA. The efficacy of temazepam for improving daytime sleep and nighttime performance in army aviators, USAARL Report, No. 2002-05, Fort Rucker, AL: US Army Aeromedical Research Laboratory, 2001. Caldwell JA, Hall KK, Erickson BS. EEG data collected from helicopter pilots in flight are sufficiently sensitive to detect increased fatigue from sleep deprivation. Int J Aviat Psychol 2002;12(1):19–32. Cole RJ. Postural baroreflex stimuli may affect EEG arousal and sleep in humans. J Appl Phys 1989;67(6):2369–2375. Dijkman M, Sachs N, Levine E, Mallis M, Carlin MM, Gillen KA, Powell JW, Samuel S, Mullington J, Rosekind MR, Dinges D. Effects of reduced stimulation on neurobehavioral alertness depend on circadian phase during human sleep deprivation. Sleep Res 1997;26:265. Dinges DF. An overview of sleepiness and accidents. J Sleep Res 1995;4(Suppl 2):4–14. Dinges DF, Pack F, Williams K, Gillen KA, Powell JW, Ott GE, Aptowicz C, Pack AI. Cumulative sleepiness, mood disturbance, and psychomotor vigilance performance decrements during a week of sleep restricted to 4-5 hours per night. Sleep 1997;20(4):267–277. Goldstein L, Murphree HB, Pfeiffer CC. Comparative study of EEG effects of antihistamines in normal volunteers. J Clin Pharm 1968;Jan/Feb:42– 53. Hickler RB, Wells RE, Tyler HR, Hamlin JT. Plasma catecholamine and electroencephalographic responses to acute postural change. Am J Med 1959;26:410–422. Kleitman N, Doktorsky A. Studies on the physiology of sleep VII. The effect of the position of the body and of sleep on rectal temperature in man. Am J Physiol 1933;104:340–343. Kra¨ uchi K, Cajochen C, Wirz-Justice A. A relationship between heat loss and sleepiness: effects of postural change and melatonin administration. J Appl Physiol 1997;83(1):134–139. Nicholson AN, Stone BM. Influence of back angle on the quality of sleep in seats. Ergonomics 1987;30(7):1033–1041. Pigeau RA, Heslegrave RJ, Angus RG. Psychophysiological measures of
J.A. Caldwell et al. / Clinical Neurophysiology 114 (2003) 23–31 drowsiness as estimators of mental fatigue and performance degradation during sleep deprivation. Electric and magnetic activity of the central nervous system: research and clinical applications in aerospace medicine, AGARD CP-432, Neuilly sur Seine: Advisory Group for Aerospace Research and Development, 1987. pp. 21-1–21-16. Ray WJ. The electrocortical system. In: Cacioppo JT, Tassinary LG, editors. Principles of psychophysiology, New York: Cambridge University Press, 1990. pp. 385–412. Sterman MB, Schummer GJ, Dushenko TW, Smith JC. Electroencephalographic correlates of pilot performance: simulation and in-flight studies. Electrical and magnetic activity of the central nervous system: research and clinical applications in aerospace medicine, AGARD CP-432,
31
Neuilly sur Seine: North Atlantic Treaty Organization, 1987. pp. 311–31-16. Sundin T. The influence of body position on the urinary excretion of adrenaline and noradrenaline. Acta Med Scand (Suppl) 1956;313:1–10. Tikuisis P, Ducharme MB. The effect of postural changes on body temperatures and heat balance. Eur J Appl Physiol 1996;72:451–455. Wilson GF, Hankins T. EEG and subjective measures of private pilot workload. Proceedings of the Human Factors and Ergonomics Society 38th annual meetingSanta Monica, CA: Human Factors and Ergonomics Society, 1994. pp. 1322–1325. Winer BJ. Statistical principles in experimental design, New York: McGraw-Hill, 1962.