- Email: [email protected]
Phase-shifting human circadian rhythms with exercise during the night shift
Physiology & Behavior, Vol. 58, No. 6, pp. 1287-1291, 1995 Copyright © 1995 Elsevier Science Inc. Printed in the USA. All rights reserved 0031-9384/95 $9.50 + .00
Pergamon 0031-9384(95)02031-P
RAPID COMMUNICATION
Phase-Shifting Human Circadian Rhythms With Exercise During the Night Shift C H A R M A N E I. E A S T M A N , 1 ERIN K. HOESE, 2 S H A W N D. Y O U N G S T E D T 2 A N D L I W E N LIU
Biological Rhythms Research Laboratory, Rush-Presbyterian-St. Luke's Medical Center, 1653 W. Congress Pkwy, Chicago, IL 60612 USA, E-Mail: [email protected] Received 3 February 1995 EASTMAN, C. I., E. K. HOESE, S. D. YOUNGSTEDT AND L. LIU. Phase-shifting human circadian rhythms with exercise during the night shift. PHYSIOL BEHAV 58(6) 1287-1291, 1995.--Appropriately timed exercise can phase shift the circadi~m rhythms of rodents. The purpose of this study was to determine whether exercise during the night shift could phase delay the temperature rhythm of humans to align with a daytime sleep schedule. Exercise subjects (N = 8) rode a ,;tationary cycle ergometer for 15 min every h during the first 3 of 8 consecutive night shifts, whereas control subjects (N = 8) remained sedentary. All subjects wore dark welder's goggles when outside after the night shift until bedtime, and then slept in dark bedrooms. Sleep was delayed 9 h from baseline. Rectal temperature was continuously measured. There were fewer evening-types and more morning-types in the exercise group than in the control group, which should have made phase delay shifts more difficult for the exercise group. Nevertheless, a majority of the exercise subjects (63%) had large temperature rhythm phase delay shifts ( > 6 h in the last 4 days relative to baseline), whereas only 38% of the control subjects had large shifts. An ANCOVA showed that, when morningness-ew~ningness was accounted for (as the covariate), the exercise group had a significantly larger temperature rhy'thm phase shift than the control group. As expected, there was a correlation between the temperature rhythm phase shift and morningness-eveningness in the control group, with greater eveningness resulting in larger phase shifts. However, there was no such relationship in the exercise group; exercise facilitated temperature rhythm phase shifts regardless of circadian type. These results suggest that exercise might be used to promote circadian adaptation to night shift work. Circadian rhythms
Exercise
Shift-work
Body temperature
INTRODUCTION
3-h bout of wheel running timed to occur during the phase advance portion of the exercise PRC (16). In contrast to the large animal literature, there has been relatively little work on the phase-shifting effects of activity on human circadian rhythms. However, as early as 1980, feedback from the activity rhythm to the circadian oscillator was proposed in a single-oscillator model of spontaneous internal desynchronization. An unusually timed sleep and wake episode (relative to the temperature rhythm) caused a small phase shift of the oscillator (2,3). Recently, direct experimental evidence suggesting a phase shifting effect of activity has been presented (22). Nocturnal exercise on a stationary bicycle for a 3-h period within an
IT IS well established that the circadian rhythms of rodents can be phase-shifted and enlxained by bouts of physical activity or exercise (8,11,15,18,19,23). Phase response curves (PRCs) for bouts of wheel-running, and various nonphotic stimuli that induce activity (such as social interaction) have been generated (15,18,21). Their basic shape shows a phase advance portion during the subjective day (when these animals are usually inactive), and a phase delay portion during the subjective night (when the animals are usually active). Activity bouts can also accelerate the rate at which circadian rhythms phase shift to align with a shift in the light-dark (LD) cycle. Reentrainment to an 8-h advance of the LD cycle was much faster when animals had one
1 TO w h o m requests for reprints should be addressed. 2 Current address: Circadian Pacemaker Laboratory, University of California, San Diego, 9500 Gilman Drive, La JoUa, CA, 92093.
1287
1288
EASTMAN ET AL.
otherwise constant routine produced phase shifts in the rhythms of melatonin and thyrotropin of 1-2 h in one day. The midpoint of the exercise bout varied from about 5 h before to 4 h after the body temperature minimum. Exercise at all time points produced phase delays, which is consistent with the exercise PRCs for nocturnal animals, in that delays were also produced during the subjective night. In the human study, there was a trend for smaller phase delays at later times, suggesting that the PRC crossover point to advances might occur later in the day. The authors noted that examining phase on the 2nd and 3rd days after exercise will be needed to determine whether the phase shifts truly reflect circadian resetting (i.e., not mere transients). A preliminary study of runners (17) found advances after morning exercise (06:00-08:00 h) compared to after evening exercise (17:00-19:00 h). It appears that these data are not consistent with the constant routine study, because 06:00-08:00 h exercise produced an advance rather than a delay. However, a comparison is difficult because no information was provided regarding the baseline temperature minima of the runners or whether their rhythms were advanced relative to baseline as well as relative to the evening condition. Furthermore, the results may have been confounded by sunlight exposure, if the morning runners were exposed to more morning (advancing) light. The circadian rhythms of shift-workers do not usually phase shift to adapt to their night-work, day-sleep schedules, but phase-shifts can be produced in some shift-work laboratory conditions (for reviews see 4,5). Phase delay shifts of circadian rhythms are also more likely in evening-types compared to morning-types (12,13). There have been two preliminary reports exploring the effects of exercise during night shift work. In one study (14), subjects exercised between 3:00 and 3:45 during 10 simulated night shifts. This exercise appeared to keep the rhythms of 3 "indifferent" or normal-types from phase delaying during the weeks of night work. In contrast, exercise had no effect on the temperature rhythms of 5 evening-types, since their rhythms gradually phase delayed regardless of whether they exercised. In the other study (20), 4 shift-workers exercised for 90 min between 21:30 and 2:30 on the first in a series of night shifts. Their temperature rhythm on subsequent night shifts was phase delayed more when they had exercised compared to when they had not exercised. These studies need to be replicated and presented in more detail before we can interpret their results with any certainty. In summary, there is some evidence that human circadian rhythms can be phase delayed by exercise during the subjective night. The purpose of this study was to determine whether moderate exercise could be used to facilitate circadian adaptation to a night-work, day-sleep schedule. The sleep period was delayed 9 h to accommodate 8 consecutive simulated night shifts. One group of subjects exercised during the first 3 night shifts, while another remained sedentary. It was hypothesized that exercise would facilitate phase delay shifts of the circadian temperature rhythm. METHOD
Subjects and General Procedures There were 16 normal subjects (10 men, 6 women), ranging in age from 19 to 41 (median = 26 yr). They participated in a simulated night shift study consisting of 7 days of baseline (with nighttime sleep) followed by 8 night shifts (with daytime sleep). All sleep periods were 8 h, with fixed bed times and wake times. The subjects slept at home in bedrooms that we made very dark by covering the windows with black plastic. Subjects were
required to remain in bed, in the dark for the entire 8 h, even if they could not sleep. The night shifts were 8 h in duration and occurred at the same clock time as the baseline sleep period. Subjects went to bed exactly 1 h after the end of each night shift, thus, the sleep/dark period was phase delayed 9 h from baseline. The first 3 night shifts were spent in the laboratory, and the remaining night shifts were spent at home. During all night shifts, the subjects remained indoors, in ordinary room light ( < 500 lx). Subjects were required to go outside for at least 5 min during the hour between the end of the night shift and bed, to simulate the minimum amount of time real shift workers might spend outdoors on their way home from night work (e.g., walking to and from a train). In our subjects, the duration of sunlight exposure was typically 5-15 min per day. Since light at this time of day will coincide with the phase advance portion of the light PRC, and thus might interfere with the phase delaying of circadian rhythms (7), subjects were required to wear dark welder's goggles (#5 lenses, about 1% transmission, Cricket frames, Uvex Safety Inc., Smithfield, RI) when outside during this" travel-home window." Subjects completed daily sleep logs to provide estimates of bedtime, sleep onset, awakenings during sleep, and final wake time. Mood was assessed twice daily, using the Profile of Mood States (POMS, 10). Each POMS rating covered one half of the waking day, and ratings were averaged for each day. Subjects completed a Momingness-Eveningness Questionnaire (9) during baseline. Several procedures were used to monitor subjects and encourage compliance with the protocol. Subjects were required to call the laboratory telephone answering machine every day at bedtime, wake time, 1 / 2 h after wake time, and every 2 h during the at-home night shifts. The answering machine recorded the date and time of each call. Subjects wore photosensors from the beginning of each night shift until bedtime, and completed daily light logs. They completed daily event logs that included entries for the consumption of caffeinated beverages. They were only allowed caffeinated beverages within the first 4 h after wake-up time, and they were not permitted to drink alcohol or take recreational drugs during the study. The daily logs and body temperature data were reviewed every few days, which helped monitor compliance with the protocol.
Exercise and Control Conditions During the first 3 night shifts, groups of 2-3 subjects sat around a table playing games. The control subjects (N = 8) were relatively sedentary except for 5 rain breaks every hour when they were permitted to walk around. The exercise subjects (N = 8) took turns on a stationary cycle ergometer (Monarch Model 818E) situated next to the table. They cycled 15 min at 50 to 60% of their maximum heart rate (HRmax) during each hour of the laboratory night shifts (i.e., 8 times per night shift during the first 3 night shifts). They spent an average of 12.3 min ( _ 0.7 SD) in their target zone. Exercise intensity was tailored to each individual's capacity by measuring HRmax with a maximal cycle ergometer test to voluntary exhaustion. These tests were conducted during the baseline week in the University of Chicago Cardiac Exercise Physiology Laboratory using the same model ergometer that was used during the study. Neither the exercise nor the control subjects were permitted to exercise outside of the lab during the last 8 days of the study.
Body Temperature Recording and Data Analysis Core body temperature was continuously measured using a flexible, disposable rectal probe and a Consumer Sensory Prod-
EXERCISE FOR PHASE-SHIFTING
1289
ucts AMS-1000 portable monitor programmed to store measurements every minute. The probes were inserted and maintained a constant distance of 10 cm. Temperature data we:re "demasked," to counteract masking by lying down and sleeping, by adding a demasking factor (a constant) to all values recorded during the planned in-bed times (cf., 1,4,6,7,24). The demasking factor was tailored to each individual on the basis of his/her temperature rhythm amplitude during baseline, and calculated as follows: First the temperature values for the last 5 days of baseline were averaged into 60 min bins. Then, the range, o:r difference between the maximum and minimum 60 min bin was determined for each day. The demasking factor was 20% of the average range. Estimates of daily circadian temperature minima (Tmin) were made by a fitting a curve composed of the 24-h fundamental cosine plus the 12-h harmonic to each 24-h section of demasked data. Two average Tmin were calculated for each subject, one for the last 5 days of baseline (days 3-7) and one for the last 4 days of the night shift (days 12-15). The difference between these two average Tmin provided a:a overall measure of temperature rhythm phase shift for each subject. Difference from baseline scores were calculated for the POMS scales and for sleep duration for each subject (the difference between the last 5 days of baseline and the last 4 days of the night shift).
night shift stand out dearly (Fig. la, bottom). For the control subject, body temperature during the night shift was relatively constant (Fig. lb, bottom). For both subjects, an increase in temperature is seen during the travel-home window between 8:00 and 9:00 (bottom graphs). Examples of the daily course of temperature phase are shown in Fig. 2. These charts of demasked data are designed to reveal the endogenous component of the temperature rhythm. The temperature rhythm of the subject on the top (a control subject) did not phase shift, whereas the temperature rhythm of the subject on the bottom (an exercise subject) phase delayed to align with the daytime sleep schedule. The phase shift of the temperature rhythm in the last 4 days relative to baseline (the difference between the average Tmin for the last 4 days of the night shift and the average Tmin for the last 5 days of baseline) was 1.3 h in the control subject and 6.9 h in the exercise subject. Summary statistics revealed that the temperature rhythm phase shift (last 4 days relative to baseline) was larger for the exercise group than the control group, 6.6 + 2.5 h vs. 4.2 + 3.4 h (mean + SD), and this difference was at the border of statistical significance by a t-test (t(14) = 1.62, p = 0.06, 1-tailed). An individual's phase shift was defined as "large" if it was > 6 h (i.e., 2 / 3 of the sleep schedule shift). By this criterion, 63% ( 5 / 8 ) of the exercise subjects had large phase shifts compared to only 38% (3/8) of the control subjects. Figure 3 shows that there was a significant negative correlation between temperature rhythm phase shift and mominguesseveningness (M-E) in the control group (r = - . 6 9 , p < 0.05, 1-tailed). In other words, greater eveningness was associated with greater phase shifts. However, this relationship was not found in the exercise group (r = -.02); phase-shifts did not depend on
RESULTS
Several aspects of the design are revealed in Fig. 1, which shows temperature wawfforms during baseline and during the first 3 night shifts for an exercise and a control subject. The increase in body temperature during the 8 exercise bouts of the
T
38
38
p r
37
I
-
i
--
137-
U r e
36 22 T
't'''i' 02 06
'I'''I'''I''' 10 14 18
36 22
38
'~'''I'
'~'''~'''I'''
22
02
06
10
14
18
22
22
02
06
10
14
18
22
38
e m
r
i ~ ) 36
. . . . . 22
02
36
06
10
Time of Day
A
14 (hr)
18
22
Time of Day (hr)
B
FIG. i. Temperature curves from raw data (not demaskcd). (A) Exercise subjectE04. (B) Control subjectEl2. Top: Average of last3 days of baseline. Verticallinesshow time of 8 h sleep/dark period (from 00:00 to 08:00).Bottom: Average of first3 days of the night shift.Verticallinesshow time of 8 h night shift(00:00-08:00) and 8 h daythne sleep/dark period (09:00-17:00). After averaging the temperature curves were smoothed by a 13 rain
moving window.
1290
EASTMAN ET AL.
morningness-eveningness. Exercise appeared to facilitate phase shifts regardless of circadian type. Figure 3 also shows that the distribution of M-E scores was slightly different in the exercise group compared to the control group, with the control group tending towards more eveningness. In the exercise group there was 1 evening type (score < 42) and 2 morning types (score > 58), whereas in the control group there were 3 evening types and 1 morning type. To determine whether this difference in morningness-eveningness was decreasing the significance of the effect of exercise, we performed a oneway analysis of covariance (ANCOVA) with M-E scores as the covariate. This analysis showed a statistically significant difference between the exercise and control groups in temperature rhythm phase shift [F(1,13) = 4.809, p = 0.047]. In other words, when differences in morningness-eveningness were accounted for, the exercise group had a significantly larger circadian rhythm phase shift than the control group. As expected, morningness-eveningnesswas related to baseline temperature phase. M-E scores were significantly correlated with the average baseline Tmin (averaged over the last 5 days of baseline) (r = - . 6 1 , df = 14, p < 0.01, 1-tailed). In other words, later Tmins were associated with greater eveningness. Since baseline temperature phase and momingness-eveningness were correlated, there were relationships between baseline phase and subsequent phase shift which were similar to those between momingness-eveningness and phase shift. Correlations were performed between temperature rhythm phase shifts (last 4 days relative to baseline) and changes in sleep
Ell
H O U R S
(a)
oo O0
. . . . . . .
:
06 '
12 . . . . .
-
' . . . . .
18 ' . . . . .
I---_-
DEMASKED
oo ~-.
I
06 '
~ '--.w..2-.i
.
12 '.
'
'
-~
~ '
9
~
•
6
•
•
• ._ rt~ .~m a. ~ '~ fig ~ ~
3
•
•
•
i
•
0
r =-.02
-3
~ [ 40
30
, 50
[
, 60
[ 70
Control 12 T .
.
.
.
.
.
.
.
l=~w! i~-~(I • |
9
i.~
tO
18 "
Exercise 12
oo
(,~
~ '---..~.~
I-.-.
--
.
30 40~ 50 60 70 Morningness-Eveningness Score 2¢
(b)
00 oo
DEMASKED
HOURS
E08 06
12 18 00 ,,,,~ ..... i ....
06
12
18
00
05-
FIG. 3. Scatter plot showing the relationship between the temperature rhythm phase shift (last 4 days relative to baseline) and the MorningnessEveningness Score. Each point represents a different subject. Lines on the x-axes mark the division between evening-types, neither-types, and morning-types. Regression lines show that phase shifts were larger with increasing eveningness in subjects who did not exercise (the control group), but that phase shifts were not constrained by morningnesseveningness in subjects who exercised.
D A ¥
10-
S 15
FIG. 2. Temperature charts from demasked data. (a) Control subject Ell. (b) Exercise subject E08. Horizontal lines show when temperature was below the daily mean. Vertical lines show the sleep/dark times, from 0:00 to 8:00 and 9:00 to 17:00. Large rectangles surrounding days 8-15 and from 0:00 to 8:00 show the time of the simulated night shifts. Shaded rectangle shows the time of exercise. Triangles show the fitted demasked temperature minima for the last 5 days of baseline and the last 4 days of the night shift.
duration and POMS scales (difference between last 4 days and baseline). As expected, larger temperature rhythm phase shifts were associated with more sleep (r = + .71, p < 0.01, 1-tailed), less Fatigue/Inertia ( r = - . 4 3 , p < 0 . 0 5 , 1-tailed), more Vigor/Activity ( r = +.38) and less Total Mood Disturbance (r = - . 3 6 ) . The last two correlation coefficients did not quite reach statistical significance with this small sample size (n = 16). DISCUSSION
These results suggest that exercise during the night shift can facilitate circadian rhythm adaptation to night work. Exercise facilitated temperature rhythm phase delays, so that the tempera-
EXERCISE FOR PHASF',-SHIFTING
1291
ture rhythm could assume a more normal phase relationship relative to the daytime sleep episodes. This and other simulated night work studies (e.g., 1,6,7) have shown that the more the temperature rhythm phase shifts to realign with sleep, the greater the improvement in sleep, alertness and mood. One factor that can account for the lack of circadian adaptation in most real night shift workers is exposure to the natural LD cycle. Sunlight exposure during the travel-home window after the night shift can coincide with the phase advance portion of the light PRC and keep circadian rhythms from delaying (7). In the present study, all subject,,; wore dark welder's goggles during the travel-home window. Another factor that prevents circadian adaptation in real night shift workers is that they do not maintain the daytime sleep schedule; instead they revert to sleeping during conventional night time laours during days off. Thus, they do not have a consistent s l e e p \ d a r k period. Furthermore, they may be exposed to natural light during the day which can interfere with phase-shifting. In this study, subjects were required to maintain a consistent daytime sleep \ d a r k schedule for 8 consecutive days. This schedule plus the dark goggles may account for the fact that a few subjects in the control group showed large temperature rhythm phase shifts, and llhese factors probably contributed to the phase shifts in the exercise group as well. Morningness-eveningness (circadian type) was also an important determinant of temperature rhythm phase shifting in the control subjects, with ewmingness predicting larger phase shifts. However, there was no relationship between momingnesseveningness and temperature rhythm phase shifts in the exercise
group; exercise produced large phase shifts regardless of circadian type. This suggests that exercise was a more powerful determinant of phase shift than circadian type. Furthermore, it suggests that exercise might be a useful technique to help morning types adapt to night work. Our data also suggest that the control of light with goggles and sleep \ dark schedules might be enough to shift the circadian rhythms of evening-type shift workers. The results of this study should be considered preliminary because of the small sample sizes. Future studies should also test exercise of different intensities, different durations and at different times of day. Interactions between bright light treatments for phase shifting and exercise should be studied. It is possible that exercise could be combined with bright light treatments for shift work, or used in shift-work situations in which bright light exposure is not possible. If future studies confirm our results, then exercise might also have other applications besides shiftwork, such as for jet-lag or circadian-rhythm-based sleep disorders (e.g., delayed sleep phase syndrome). Appropriately timed exercise might also be used to help entrain free-running blind people. ACKNOWLEDGEMENTS We thank Louis F. Fogg, Ph.D. for statistical advice, and Kristienne Kattapong and Robert Tell for technical assistance. Goggles were donated by Uvex Safety, Inc., Smithfield, R.I. This work was supported by NIH grant NS23421.
REFERENCES
1. Dawson, D.; Campbell S. S. Timed exposure to bright light improves sleep and alerlness during simulated night shifts. Sleep 14:511-516; 1991. 2. Eastman, C. I. Are separate temperature and activity oscillators necessary to explain the phenomena of human circadian rhythms? In: Moore-Ede, M. C.; Czeisler, C. A., eds. Mathematical models of the circadian sleep-wake cycle. New York: Raven Press; 1984:81-103. 3. Eastman, C. I. Circadian rhythms of rats at the limits of entrainment and a phase-shift model of "spontaneous internal desynchronization" in humans. In: Sleep 1980. 5th Eur. Congr. Sleep Res. Amsterdam. Basel: Karger; 1980:59--64. 4. Eastman, C. I. High intensity light for circadian adaptation to a 12-h shift of the sleep schedttle. Am. J. Physiol. 263:R428-R436; 1992. 5. Eastman, C. I.; Boulos, Z.; Terman, M.; Campbell, S. S.; Dijk, D.-J.; Lewy, A. J. Light treatment for sleep disorders: Consensus report. VI. Shift work. J. Biol. Rhythms. 10:138-145; 1995. 6. Eastman, C. I.; Liu, L.; Fogg, L. F. Circadian rhythm adaptation to simulated night shift work: Effect of nocturnal bright-light duration. Sleep 18:399-407; 1995. 7. Eastman, C. I.; Stewart, K. T.; Mahoney, M. P.; Lin, L.; Fogg, L. F. Dark goggles and bright light improve circadian rhythm adaptation to night-shift work. Sleep 17:535-543; 1994. 8. Edgar, D. M.; Dement, W. C. Regularly scheduled voluntary exercise synchronizes the mouse circadian clock. Am. J. Physiol. 261:R928R933; 1991. 9. Home, A.; Ostherg, O. Self-assessment questionnaire to determine momingness-eveningness in human circadian rhythms. Int. J. Chronobiol. 4:97-110; 1976. 10. McNair, D. M.; Lorr, M.; Droppleman, L. F. Manual for the profile of mood states. San Diego: Educational and Industrial Testing Service; 1971. 11. Misflberger, R. E. Scheduled daily exercise or feeding alters the phase of photic entrainment in Syrian hamsters. Physiol. Behav. 50:1257-1260; 1991. 12. Mitchell, P. J.; Hoese, E. K.; Liu, L.; Fogg, L. F.; Eastman, C. I. Facilitating vs. conflicting bright light exposure during night shifts for circadian adaptation to delayed and advanced sleep schedule shifts. Sleep Res. 24:5Z~; 1995.
13. Moog, R. Optimization of shift work: Physiological contributions. Ergonomics 30:1249-1259; 1987. 14. Moog, R.; Hildebrandt, G. Effects of physical training on adaptation to night-work. In: Oginski, A.; Pokorski, J.; Rutenfranz, J., eds. Contemporary advances in shiftwork research: Theoretical and practical aspects in the late eighties. Krakow: Medical Academy; 1987:8185. 15. Mrosovsky, N.; Rcebs, S. G.; Honrado, G. I.; Salmon, P. A. Behavioural entrainment of circadian rhythms. Experientia 45:696-702; 1989. 16. Mrosovsky, N.; Salmon, P. A. A behavioural method for accelerating reentrainment of rhythms to new light-dark cycles. Nature 330:372373; 1987. 17. Piercy, J.; Lack, L. Daily exercise can shift the endogenous circadian phase. Sleep Res. 17:393; 1988. 18. Reebs, S. G.; Mrosovsky. Effects of induced wheel running on the circadian activity rhythms of syrian hamsters: Entrainment and phase response curve. J. Biol. Rhythms. 4:39-48; 1989. 19. Reebs, S. G.; Mrosovsky, N. Large phase-shifts of circadian rhythms caused by induced running in a reentrainment paradigm: The role of pulse duration and light. J. Comp. Physiol. A. 165:819-825; 1989. 20. Schmidt, K. P.; Koehler, W. K.; Fleissner, G.; Pflug, B. Locomotor activity accelerates the adjustment of the temperature rhythm in shiftwork. In: Diez-Noguera, A.; Cambras, T., eds. Chronobiology & Chronomedicine. New York: Peter Lang; 1992:389-395. 21. Smith, R. D.; Turek, F. W.; Takahashi, J. S. Two families of phase-response curves characterize the resetting of the hamster circadian clock. Am. J. Physiol. 262:Rl149-Rl153; 1992. 22. van Reeth, O.; Sturis, J.; Byme, M. M.; Blackman, J. D.; L'Hermite-Baleriaux, M.; Leproult, R.; Oliner, C.; Refetoff, S.; Turek, F. W.; van Canter, E. Nocturnal exercise phase delays circadian rhythms of melatonin and thyrotropin secretion in normal men. Am. J. Physiol. 266:E964-E974; 1994. 23. van Rceth, O.; Turek, F. W. Stimulated activity mediates phase shifts in the hamster circadian clock induced by dark pulses or benzodiazepines. Nature 339:49-51; 1989. 24. Wever, R. A. The circadian system of man. New York-HeidelbergBerlin: Springer-Veflag; 1979.
Pergamon 0031-9384(95)02031-P
RAPID COMMUNICATION
Phase-Shifting Human Circadian Rhythms With Exercise During the Night Shift C H A R M A N E I. E A S T M A N , 1 ERIN K. HOESE, 2 S H A W N D. Y O U N G S T E D T 2 A N D L I W E N LIU
Biological Rhythms Research Laboratory, Rush-Presbyterian-St. Luke's Medical Center, 1653 W. Congress Pkwy, Chicago, IL 60612 USA, E-Mail: [email protected] Received 3 February 1995 EASTMAN, C. I., E. K. HOESE, S. D. YOUNGSTEDT AND L. LIU. Phase-shifting human circadian rhythms with exercise during the night shift. PHYSIOL BEHAV 58(6) 1287-1291, 1995.--Appropriately timed exercise can phase shift the circadi~m rhythms of rodents. The purpose of this study was to determine whether exercise during the night shift could phase delay the temperature rhythm of humans to align with a daytime sleep schedule. Exercise subjects (N = 8) rode a ,;tationary cycle ergometer for 15 min every h during the first 3 of 8 consecutive night shifts, whereas control subjects (N = 8) remained sedentary. All subjects wore dark welder's goggles when outside after the night shift until bedtime, and then slept in dark bedrooms. Sleep was delayed 9 h from baseline. Rectal temperature was continuously measured. There were fewer evening-types and more morning-types in the exercise group than in the control group, which should have made phase delay shifts more difficult for the exercise group. Nevertheless, a majority of the exercise subjects (63%) had large temperature rhythm phase delay shifts ( > 6 h in the last 4 days relative to baseline), whereas only 38% of the control subjects had large shifts. An ANCOVA showed that, when morningness-ew~ningness was accounted for (as the covariate), the exercise group had a significantly larger temperature rhy'thm phase shift than the control group. As expected, there was a correlation between the temperature rhythm phase shift and morningness-eveningness in the control group, with greater eveningness resulting in larger phase shifts. However, there was no such relationship in the exercise group; exercise facilitated temperature rhythm phase shifts regardless of circadian type. These results suggest that exercise might be used to promote circadian adaptation to night shift work. Circadian rhythms
Exercise
Shift-work
Body temperature
INTRODUCTION
3-h bout of wheel running timed to occur during the phase advance portion of the exercise PRC (16). In contrast to the large animal literature, there has been relatively little work on the phase-shifting effects of activity on human circadian rhythms. However, as early as 1980, feedback from the activity rhythm to the circadian oscillator was proposed in a single-oscillator model of spontaneous internal desynchronization. An unusually timed sleep and wake episode (relative to the temperature rhythm) caused a small phase shift of the oscillator (2,3). Recently, direct experimental evidence suggesting a phase shifting effect of activity has been presented (22). Nocturnal exercise on a stationary bicycle for a 3-h period within an
IT IS well established that the circadian rhythms of rodents can be phase-shifted and enlxained by bouts of physical activity or exercise (8,11,15,18,19,23). Phase response curves (PRCs) for bouts of wheel-running, and various nonphotic stimuli that induce activity (such as social interaction) have been generated (15,18,21). Their basic shape shows a phase advance portion during the subjective day (when these animals are usually inactive), and a phase delay portion during the subjective night (when the animals are usually active). Activity bouts can also accelerate the rate at which circadian rhythms phase shift to align with a shift in the light-dark (LD) cycle. Reentrainment to an 8-h advance of the LD cycle was much faster when animals had one
1 TO w h o m requests for reprints should be addressed. 2 Current address: Circadian Pacemaker Laboratory, University of California, San Diego, 9500 Gilman Drive, La JoUa, CA, 92093.
1287
1288
EASTMAN ET AL.
otherwise constant routine produced phase shifts in the rhythms of melatonin and thyrotropin of 1-2 h in one day. The midpoint of the exercise bout varied from about 5 h before to 4 h after the body temperature minimum. Exercise at all time points produced phase delays, which is consistent with the exercise PRCs for nocturnal animals, in that delays were also produced during the subjective night. In the human study, there was a trend for smaller phase delays at later times, suggesting that the PRC crossover point to advances might occur later in the day. The authors noted that examining phase on the 2nd and 3rd days after exercise will be needed to determine whether the phase shifts truly reflect circadian resetting (i.e., not mere transients). A preliminary study of runners (17) found advances after morning exercise (06:00-08:00 h) compared to after evening exercise (17:00-19:00 h). It appears that these data are not consistent with the constant routine study, because 06:00-08:00 h exercise produced an advance rather than a delay. However, a comparison is difficult because no information was provided regarding the baseline temperature minima of the runners or whether their rhythms were advanced relative to baseline as well as relative to the evening condition. Furthermore, the results may have been confounded by sunlight exposure, if the morning runners were exposed to more morning (advancing) light. The circadian rhythms of shift-workers do not usually phase shift to adapt to their night-work, day-sleep schedules, but phase-shifts can be produced in some shift-work laboratory conditions (for reviews see 4,5). Phase delay shifts of circadian rhythms are also more likely in evening-types compared to morning-types (12,13). There have been two preliminary reports exploring the effects of exercise during night shift work. In one study (14), subjects exercised between 3:00 and 3:45 during 10 simulated night shifts. This exercise appeared to keep the rhythms of 3 "indifferent" or normal-types from phase delaying during the weeks of night work. In contrast, exercise had no effect on the temperature rhythms of 5 evening-types, since their rhythms gradually phase delayed regardless of whether they exercised. In the other study (20), 4 shift-workers exercised for 90 min between 21:30 and 2:30 on the first in a series of night shifts. Their temperature rhythm on subsequent night shifts was phase delayed more when they had exercised compared to when they had not exercised. These studies need to be replicated and presented in more detail before we can interpret their results with any certainty. In summary, there is some evidence that human circadian rhythms can be phase delayed by exercise during the subjective night. The purpose of this study was to determine whether moderate exercise could be used to facilitate circadian adaptation to a night-work, day-sleep schedule. The sleep period was delayed 9 h to accommodate 8 consecutive simulated night shifts. One group of subjects exercised during the first 3 night shifts, while another remained sedentary. It was hypothesized that exercise would facilitate phase delay shifts of the circadian temperature rhythm. METHOD
Subjects and General Procedures There were 16 normal subjects (10 men, 6 women), ranging in age from 19 to 41 (median = 26 yr). They participated in a simulated night shift study consisting of 7 days of baseline (with nighttime sleep) followed by 8 night shifts (with daytime sleep). All sleep periods were 8 h, with fixed bed times and wake times. The subjects slept at home in bedrooms that we made very dark by covering the windows with black plastic. Subjects were
required to remain in bed, in the dark for the entire 8 h, even if they could not sleep. The night shifts were 8 h in duration and occurred at the same clock time as the baseline sleep period. Subjects went to bed exactly 1 h after the end of each night shift, thus, the sleep/dark period was phase delayed 9 h from baseline. The first 3 night shifts were spent in the laboratory, and the remaining night shifts were spent at home. During all night shifts, the subjects remained indoors, in ordinary room light ( < 500 lx). Subjects were required to go outside for at least 5 min during the hour between the end of the night shift and bed, to simulate the minimum amount of time real shift workers might spend outdoors on their way home from night work (e.g., walking to and from a train). In our subjects, the duration of sunlight exposure was typically 5-15 min per day. Since light at this time of day will coincide with the phase advance portion of the light PRC, and thus might interfere with the phase delaying of circadian rhythms (7), subjects were required to wear dark welder's goggles (#5 lenses, about 1% transmission, Cricket frames, Uvex Safety Inc., Smithfield, RI) when outside during this" travel-home window." Subjects completed daily sleep logs to provide estimates of bedtime, sleep onset, awakenings during sleep, and final wake time. Mood was assessed twice daily, using the Profile of Mood States (POMS, 10). Each POMS rating covered one half of the waking day, and ratings were averaged for each day. Subjects completed a Momingness-Eveningness Questionnaire (9) during baseline. Several procedures were used to monitor subjects and encourage compliance with the protocol. Subjects were required to call the laboratory telephone answering machine every day at bedtime, wake time, 1 / 2 h after wake time, and every 2 h during the at-home night shifts. The answering machine recorded the date and time of each call. Subjects wore photosensors from the beginning of each night shift until bedtime, and completed daily light logs. They completed daily event logs that included entries for the consumption of caffeinated beverages. They were only allowed caffeinated beverages within the first 4 h after wake-up time, and they were not permitted to drink alcohol or take recreational drugs during the study. The daily logs and body temperature data were reviewed every few days, which helped monitor compliance with the protocol.
Exercise and Control Conditions During the first 3 night shifts, groups of 2-3 subjects sat around a table playing games. The control subjects (N = 8) were relatively sedentary except for 5 rain breaks every hour when they were permitted to walk around. The exercise subjects (N = 8) took turns on a stationary cycle ergometer (Monarch Model 818E) situated next to the table. They cycled 15 min at 50 to 60% of their maximum heart rate (HRmax) during each hour of the laboratory night shifts (i.e., 8 times per night shift during the first 3 night shifts). They spent an average of 12.3 min ( _ 0.7 SD) in their target zone. Exercise intensity was tailored to each individual's capacity by measuring HRmax with a maximal cycle ergometer test to voluntary exhaustion. These tests were conducted during the baseline week in the University of Chicago Cardiac Exercise Physiology Laboratory using the same model ergometer that was used during the study. Neither the exercise nor the control subjects were permitted to exercise outside of the lab during the last 8 days of the study.
Body Temperature Recording and Data Analysis Core body temperature was continuously measured using a flexible, disposable rectal probe and a Consumer Sensory Prod-
EXERCISE FOR PHASE-SHIFTING
1289
ucts AMS-1000 portable monitor programmed to store measurements every minute. The probes were inserted and maintained a constant distance of 10 cm. Temperature data we:re "demasked," to counteract masking by lying down and sleeping, by adding a demasking factor (a constant) to all values recorded during the planned in-bed times (cf., 1,4,6,7,24). The demasking factor was tailored to each individual on the basis of his/her temperature rhythm amplitude during baseline, and calculated as follows: First the temperature values for the last 5 days of baseline were averaged into 60 min bins. Then, the range, o:r difference between the maximum and minimum 60 min bin was determined for each day. The demasking factor was 20% of the average range. Estimates of daily circadian temperature minima (Tmin) were made by a fitting a curve composed of the 24-h fundamental cosine plus the 12-h harmonic to each 24-h section of demasked data. Two average Tmin were calculated for each subject, one for the last 5 days of baseline (days 3-7) and one for the last 4 days of the night shift (days 12-15). The difference between these two average Tmin provided a:a overall measure of temperature rhythm phase shift for each subject. Difference from baseline scores were calculated for the POMS scales and for sleep duration for each subject (the difference between the last 5 days of baseline and the last 4 days of the night shift).
night shift stand out dearly (Fig. la, bottom). For the control subject, body temperature during the night shift was relatively constant (Fig. lb, bottom). For both subjects, an increase in temperature is seen during the travel-home window between 8:00 and 9:00 (bottom graphs). Examples of the daily course of temperature phase are shown in Fig. 2. These charts of demasked data are designed to reveal the endogenous component of the temperature rhythm. The temperature rhythm of the subject on the top (a control subject) did not phase shift, whereas the temperature rhythm of the subject on the bottom (an exercise subject) phase delayed to align with the daytime sleep schedule. The phase shift of the temperature rhythm in the last 4 days relative to baseline (the difference between the average Tmin for the last 4 days of the night shift and the average Tmin for the last 5 days of baseline) was 1.3 h in the control subject and 6.9 h in the exercise subject. Summary statistics revealed that the temperature rhythm phase shift (last 4 days relative to baseline) was larger for the exercise group than the control group, 6.6 + 2.5 h vs. 4.2 + 3.4 h (mean + SD), and this difference was at the border of statistical significance by a t-test (t(14) = 1.62, p = 0.06, 1-tailed). An individual's phase shift was defined as "large" if it was > 6 h (i.e., 2 / 3 of the sleep schedule shift). By this criterion, 63% ( 5 / 8 ) of the exercise subjects had large phase shifts compared to only 38% (3/8) of the control subjects. Figure 3 shows that there was a significant negative correlation between temperature rhythm phase shift and mominguesseveningness (M-E) in the control group (r = - . 6 9 , p < 0.05, 1-tailed). In other words, greater eveningness was associated with greater phase shifts. However, this relationship was not found in the exercise group (r = -.02); phase-shifts did not depend on
RESULTS
Several aspects of the design are revealed in Fig. 1, which shows temperature wawfforms during baseline and during the first 3 night shifts for an exercise and a control subject. The increase in body temperature during the 8 exercise bouts of the
T
38
38
p r
37
I
-
i
--
137-
U r e
36 22 T
't'''i' 02 06
'I'''I'''I''' 10 14 18
36 22
38
'~'''I'
'~'''~'''I'''
22
02
06
10
14
18
22
22
02
06
10
14
18
22
38
e m
r
i ~ ) 36
. . . . . 22
02
36
06
10
Time of Day
A
14 (hr)
18
22
Time of Day (hr)
B
FIG. i. Temperature curves from raw data (not demaskcd). (A) Exercise subjectE04. (B) Control subjectEl2. Top: Average of last3 days of baseline. Verticallinesshow time of 8 h sleep/dark period (from 00:00 to 08:00).Bottom: Average of first3 days of the night shift.Verticallinesshow time of 8 h night shift(00:00-08:00) and 8 h daythne sleep/dark period (09:00-17:00). After averaging the temperature curves were smoothed by a 13 rain
moving window.
1290
EASTMAN ET AL.
morningness-eveningness. Exercise appeared to facilitate phase shifts regardless of circadian type. Figure 3 also shows that the distribution of M-E scores was slightly different in the exercise group compared to the control group, with the control group tending towards more eveningness. In the exercise group there was 1 evening type (score < 42) and 2 morning types (score > 58), whereas in the control group there were 3 evening types and 1 morning type. To determine whether this difference in morningness-eveningness was decreasing the significance of the effect of exercise, we performed a oneway analysis of covariance (ANCOVA) with M-E scores as the covariate. This analysis showed a statistically significant difference between the exercise and control groups in temperature rhythm phase shift [F(1,13) = 4.809, p = 0.047]. In other words, when differences in morningness-eveningness were accounted for, the exercise group had a significantly larger circadian rhythm phase shift than the control group. As expected, morningness-eveningnesswas related to baseline temperature phase. M-E scores were significantly correlated with the average baseline Tmin (averaged over the last 5 days of baseline) (r = - . 6 1 , df = 14, p < 0.01, 1-tailed). In other words, later Tmins were associated with greater eveningness. Since baseline temperature phase and momingness-eveningness were correlated, there were relationships between baseline phase and subsequent phase shift which were similar to those between momingness-eveningness and phase shift. Correlations were performed between temperature rhythm phase shifts (last 4 days relative to baseline) and changes in sleep
Ell
H O U R S
(a)
oo O0
. . . . . . .
:
06 '
12 . . . . .
-
' . . . . .
18 ' . . . . .
I---_-
DEMASKED
oo ~-.
I
06 '
~ '--.w..2-.i
.
12 '.
'
'
-~
~ '
9
~
•
6
•
•
• ._ rt~ .~m a. ~ '~ fig ~ ~
3
•
•
•
i
•
0
r =-.02
-3
~ [ 40
30
, 50
[
, 60
[ 70
Control 12 T .
.
.
.
.
.
.
.
l=~w! i~-~(I • |
9
i.~
tO
18 "
Exercise 12
oo
(,~
~ '---..~.~
I-.-.
--
.
30 40~ 50 60 70 Morningness-Eveningness Score 2¢
(b)
00 oo
DEMASKED
HOURS
E08 06
12 18 00 ,,,,~ ..... i ....
06
12
18
00
05-
FIG. 3. Scatter plot showing the relationship between the temperature rhythm phase shift (last 4 days relative to baseline) and the MorningnessEveningness Score. Each point represents a different subject. Lines on the x-axes mark the division between evening-types, neither-types, and morning-types. Regression lines show that phase shifts were larger with increasing eveningness in subjects who did not exercise (the control group), but that phase shifts were not constrained by morningnesseveningness in subjects who exercised.
D A ¥
10-
S 15
FIG. 2. Temperature charts from demasked data. (a) Control subject Ell. (b) Exercise subject E08. Horizontal lines show when temperature was below the daily mean. Vertical lines show the sleep/dark times, from 0:00 to 8:00 and 9:00 to 17:00. Large rectangles surrounding days 8-15 and from 0:00 to 8:00 show the time of the simulated night shifts. Shaded rectangle shows the time of exercise. Triangles show the fitted demasked temperature minima for the last 5 days of baseline and the last 4 days of the night shift.
duration and POMS scales (difference between last 4 days and baseline). As expected, larger temperature rhythm phase shifts were associated with more sleep (r = + .71, p < 0.01, 1-tailed), less Fatigue/Inertia ( r = - . 4 3 , p < 0 . 0 5 , 1-tailed), more Vigor/Activity ( r = +.38) and less Total Mood Disturbance (r = - . 3 6 ) . The last two correlation coefficients did not quite reach statistical significance with this small sample size (n = 16). DISCUSSION
These results suggest that exercise during the night shift can facilitate circadian rhythm adaptation to night work. Exercise facilitated temperature rhythm phase delays, so that the tempera-
EXERCISE FOR PHASF',-SHIFTING
1291
ture rhythm could assume a more normal phase relationship relative to the daytime sleep episodes. This and other simulated night work studies (e.g., 1,6,7) have shown that the more the temperature rhythm phase shifts to realign with sleep, the greater the improvement in sleep, alertness and mood. One factor that can account for the lack of circadian adaptation in most real night shift workers is exposure to the natural LD cycle. Sunlight exposure during the travel-home window after the night shift can coincide with the phase advance portion of the light PRC and keep circadian rhythms from delaying (7). In the present study, all subject,,; wore dark welder's goggles during the travel-home window. Another factor that prevents circadian adaptation in real night shift workers is that they do not maintain the daytime sleep schedule; instead they revert to sleeping during conventional night time laours during days off. Thus, they do not have a consistent s l e e p \ d a r k period. Furthermore, they may be exposed to natural light during the day which can interfere with phase-shifting. In this study, subjects were required to maintain a consistent daytime sleep \ d a r k schedule for 8 consecutive days. This schedule plus the dark goggles may account for the fact that a few subjects in the control group showed large temperature rhythm phase shifts, and llhese factors probably contributed to the phase shifts in the exercise group as well. Morningness-eveningness (circadian type) was also an important determinant of temperature rhythm phase shifting in the control subjects, with ewmingness predicting larger phase shifts. However, there was no relationship between momingnesseveningness and temperature rhythm phase shifts in the exercise
group; exercise produced large phase shifts regardless of circadian type. This suggests that exercise was a more powerful determinant of phase shift than circadian type. Furthermore, it suggests that exercise might be a useful technique to help morning types adapt to night work. Our data also suggest that the control of light with goggles and sleep \ dark schedules might be enough to shift the circadian rhythms of evening-type shift workers. The results of this study should be considered preliminary because of the small sample sizes. Future studies should also test exercise of different intensities, different durations and at different times of day. Interactions between bright light treatments for phase shifting and exercise should be studied. It is possible that exercise could be combined with bright light treatments for shift work, or used in shift-work situations in which bright light exposure is not possible. If future studies confirm our results, then exercise might also have other applications besides shiftwork, such as for jet-lag or circadian-rhythm-based sleep disorders (e.g., delayed sleep phase syndrome). Appropriately timed exercise might also be used to help entrain free-running blind people. ACKNOWLEDGEMENTS We thank Louis F. Fogg, Ph.D. for statistical advice, and Kristienne Kattapong and Robert Tell for technical assistance. Goggles were donated by Uvex Safety, Inc., Smithfield, R.I. This work was supported by NIH grant NS23421.
REFERENCES
1. Dawson, D.; Campbell S. S. Timed exposure to bright light improves sleep and alerlness during simulated night shifts. Sleep 14:511-516; 1991. 2. Eastman, C. I. Are separate temperature and activity oscillators necessary to explain the phenomena of human circadian rhythms? In: Moore-Ede, M. C.; Czeisler, C. A., eds. Mathematical models of the circadian sleep-wake cycle. New York: Raven Press; 1984:81-103. 3. Eastman, C. I. Circadian rhythms of rats at the limits of entrainment and a phase-shift model of "spontaneous internal desynchronization" in humans. In: Sleep 1980. 5th Eur. Congr. Sleep Res. Amsterdam. Basel: Karger; 1980:59--64. 4. Eastman, C. I. High intensity light for circadian adaptation to a 12-h shift of the sleep schedttle. Am. J. Physiol. 263:R428-R436; 1992. 5. Eastman, C. I.; Boulos, Z.; Terman, M.; Campbell, S. S.; Dijk, D.-J.; Lewy, A. J. Light treatment for sleep disorders: Consensus report. VI. Shift work. J. Biol. Rhythms. 10:138-145; 1995. 6. Eastman, C. I.; Liu, L.; Fogg, L. F. Circadian rhythm adaptation to simulated night shift work: Effect of nocturnal bright-light duration. Sleep 18:399-407; 1995. 7. Eastman, C. I.; Stewart, K. T.; Mahoney, M. P.; Lin, L.; Fogg, L. F. Dark goggles and bright light improve circadian rhythm adaptation to night-shift work. Sleep 17:535-543; 1994. 8. Edgar, D. M.; Dement, W. C. Regularly scheduled voluntary exercise synchronizes the mouse circadian clock. Am. J. Physiol. 261:R928R933; 1991. 9. Home, A.; Ostherg, O. Self-assessment questionnaire to determine momingness-eveningness in human circadian rhythms. Int. J. Chronobiol. 4:97-110; 1976. 10. McNair, D. M.; Lorr, M.; Droppleman, L. F. Manual for the profile of mood states. San Diego: Educational and Industrial Testing Service; 1971. 11. Misflberger, R. E. Scheduled daily exercise or feeding alters the phase of photic entrainment in Syrian hamsters. Physiol. Behav. 50:1257-1260; 1991. 12. Mitchell, P. J.; Hoese, E. K.; Liu, L.; Fogg, L. F.; Eastman, C. I. Facilitating vs. conflicting bright light exposure during night shifts for circadian adaptation to delayed and advanced sleep schedule shifts. Sleep Res. 24:5Z~; 1995.
13. Moog, R. Optimization of shift work: Physiological contributions. Ergonomics 30:1249-1259; 1987. 14. Moog, R.; Hildebrandt, G. Effects of physical training on adaptation to night-work. In: Oginski, A.; Pokorski, J.; Rutenfranz, J., eds. Contemporary advances in shiftwork research: Theoretical and practical aspects in the late eighties. Krakow: Medical Academy; 1987:8185. 15. Mrosovsky, N.; Rcebs, S. G.; Honrado, G. I.; Salmon, P. A. Behavioural entrainment of circadian rhythms. Experientia 45:696-702; 1989. 16. Mrosovsky, N.; Salmon, P. A. A behavioural method for accelerating reentrainment of rhythms to new light-dark cycles. Nature 330:372373; 1987. 17. Piercy, J.; Lack, L. Daily exercise can shift the endogenous circadian phase. Sleep Res. 17:393; 1988. 18. Reebs, S. G.; Mrosovsky. Effects of induced wheel running on the circadian activity rhythms of syrian hamsters: Entrainment and phase response curve. J. Biol. Rhythms. 4:39-48; 1989. 19. Reebs, S. G.; Mrosovsky, N. Large phase-shifts of circadian rhythms caused by induced running in a reentrainment paradigm: The role of pulse duration and light. J. Comp. Physiol. A. 165:819-825; 1989. 20. Schmidt, K. P.; Koehler, W. K.; Fleissner, G.; Pflug, B. Locomotor activity accelerates the adjustment of the temperature rhythm in shiftwork. In: Diez-Noguera, A.; Cambras, T., eds. Chronobiology & Chronomedicine. New York: Peter Lang; 1992:389-395. 21. Smith, R. D.; Turek, F. W.; Takahashi, J. S. Two families of phase-response curves characterize the resetting of the hamster circadian clock. Am. J. Physiol. 262:Rl149-Rl153; 1992. 22. van Reeth, O.; Sturis, J.; Byme, M. M.; Blackman, J. D.; L'Hermite-Baleriaux, M.; Leproult, R.; Oliner, C.; Refetoff, S.; Turek, F. W.; van Canter, E. Nocturnal exercise phase delays circadian rhythms of melatonin and thyrotropin secretion in normal men. Am. J. Physiol. 266:E964-E974; 1994. 23. van Rceth, O.; Turek, F. W. Stimulated activity mediates phase shifts in the hamster circadian clock induced by dark pulses or benzodiazepines. Nature 339:49-51; 1989. 24. Wever, R. A. The circadian system of man. New York-HeidelbergBerlin: Springer-Veflag; 1979.