Sleep, 19(7):539-543
© 1996 American Sleep Disorders Association and Sleep Research Society
Circadian Rhythms and Sleep
Fast Track Publication
Rapid Shift in Sleep Time and Acrophase of Melatonin
Secretion in Short Shift Work Schedule
*Maria Antonia Quera-Salva, tRemy Defrance, :j:Bruno Claustrat,
*Jacques De Lattre and §Christian Guilleminault
*Unite du Sommeil, Hopital Raymond Poincarre, Garches, France;
tDivision Neurobiologie, IRIS, Courbevoie, France;
tService de Radio-pharmacie et Radio-analyse, Hopital Neuro-Cardiologique, Lyon, France; and
§Sleep Research Center, Stanford University School of Medicine, Stanford, California, U.S.A.
Summary: Tolerance to shift work and adaptability to shifting schedules is an issue of growing importance in
industrialized society. We studied 40 registered nurses, 20 on fixed day-shifts and 20 on fixed night-shifts, to assess
whether workers with rapidly shifting schedules were able to adapt their melatonin secretion and sleep-wake cycles.
The day-shift worked 5 days with 2 days off and the night-shift worked 3 nights with 2 off. All night-shift
personnel acknowledged shifting back to daytime schedules on their days off. Sleep-wake was determined by sleep
logs and actigraphy. To measure 6-sulfatoxymelatonin levels, urine was collected at 2-hour intervals on the last
work day and on the last day off.
Night-shift workers slept significantly more on days off. Napping on the job occurred in 9120 night-shift
workers (mean 114 minutes) between 3 and 6 a.m. The acrophase of 6-sulfatoxymelatonin in day-shift nurses
occurred at similar times on workdays and off days. In night-shift nurses, the acrophase was about 7 a.m. on
days off, but had a random distribution on workdays. Further analysis revealed two subgroups of night-shift
nurses: six subjects (group A) demonstrated a rapid shift in melatonin secretion (acrophase at near 12 noon on
work days and at near 7 a.m. on days off) while 14 nurses (group B) did not shift. Group A nurses slept more
in the daytime on work days and their total sleep time was the same as day-shift nurses. Group A was slightly
younger and was composed solely of women (there were nine women and five men in group B). Age may be a
factor in the ability to adapt to rapidly shifting schedules. Key Words: Shift work-Melatonin-SIeep-wake
schedule-Fast day/night shift.
Ten percent of the adult population of the European
community engages in shift work, as does one in four
adult men and one in six adult women in the U.S.A.
A consequence of the industrial revolution, shift work
poses a wide array of health problems (1). The investigation of circadian rhythms has taken on great importance as we strive to understand the pathologies
associated with societally induced dyschronosis. Melatonin has been recognized as an important marker of
circadian regulation, with peak secretion usually ocAccepted for publication May 1996.
Address correspondence and reprint requests to Dr. Quera-Salva,
Unite du Sommeil, Hopital Raymond Poincarre, Boulevard Raymond Poincarre, Garches, France.
curring during nocturnal sleep between 2 and 4 a.m,
(1). It has been suggested that synchronizing melatonin
secretion with the desired sleep-wake schedule could
lead to an increased feeling of well-being for shift
workers (2,3). Preliminary information from Sack et
al. (4) and Roden et al. (3), however, was obtained on
small groups of subjects and provided conflicting results. In this study, we measured the urinary excretion
of the main melatonin metabolite-6-sulfatoxymelatonin-in day-shift workers and night-shift workers
who reverted to a daytime schedule on their days off.
The goal of our study was to determine whether workers who shift rapidly from day-wake to night-wake
schedules were able to adapt their melatonin secretion
and sleep-wake cycles appropriately.
539
M. A. QUERA-SALVA ET AL.
540
METHODS
TABLE 1.
Sociodemographic and clinical information
Population
The studied population included 40 registered nurses, 20 on a fixed night-shift and 20 on a fixed dayshift. The nurses worked in the same hospital and in
the same departments, and were matched for age, gender, and, as much as possible, sociofamilial responsibilities. To participate in the study, the nurses had to
sign informed consent forms approved by the institution's review board; be between 25 and 55 years of
age; be free of significant medical or psychiatric illnesses; be in generally good health at the beginning
of the study; take no drugs that could affect melatonin
production; and have anxiety, depression, and "global" scores on the symptom checklist below 65 (5).
Subjects must also have been assigned to their permanent schedule, without a single absence for any reason, for at least 3 months prior to the study. All subjects were studied during the late spring and early summer of 1994.
Work schedules
The day-shift group worked 5 successive days then
had 2 days off for the 3 months leading up to and
during the study period. These nurses had a 39-hour
week and worked from 0700 to 1500 hours. The nightshift group worked 3 days on and 2 days off during
the experimental study period. These nurses had a
35-hour work week, and worked from 2100 to 0700
hours. In the 15 days leading up to the investigation,
their work schedule was as follows: 3 days on, 2 days
off; 2 days on, 3 days off; 2 days on, 3 days off. All
night-shift personnel acknowledged shifting back to
daytime schedules on off days.
Procedures
At the time of recruitment, subjects were given a
brief clinical evaluation with a basic blood and urine
analysis. They also were asked to fill out the "self
assessment questionnaire to determine momingness
and eveningness in human circadian rhythm" (6), the
European Standard Shift Work Index Questionnaire
(7), and the Derogatis Symptoms Check List (5). Subjects were asked to fill out sleep/wake diaries beginning 15 days prior to the start of the study and
throughout the study period.
The experimental period lasted 7 days for day-shift
workers (5 workdays and 2 off days), and 5 days for
night-shift workers (3 workdays and 2 off days). Each
subject was asked to wear a wrist actigraph (Gaehwiller electronic, Physiocom, France) on the nondomSleep, Vol. 19, No.7, 1996
Gender
Age (years)
BMI (kg/m2)
Years of nursing
Years with current schedule
Married or living with partner
Divorced/separated
Single
Number of people to take care of
0
I, 2
23
Happy with schedule
Very
Somewhat
Not
Night
workers
(n = 20)
Day
workers
(n = 20)
4 M/16 F
36:!: 7
24:!: 2
13 :!: 9
6:!:4
13
2
5
at home
5
11
4
4 Ml16 F
35 :!: 7
22:!: 2
12 ~ 7
7:!:5
14
1
5
15
1
4
Significance
ns
ns
0.04
ns
ns
7
12
1
18
0
2
inant arm throughout the experimental period. The actigraphs, which monitor activity/inactivity and whose
results have been correlated with wake and sleep in
several previous reports (8,9,10), were preset with a
sampling rate of one data point per minute.
Urine was collected from all subjects at 2-hour intervals during two 24-hour periods. The first series of
urine collections was performed on the last workday
of the experimental period (day 5 for day-shift nurses
and "day" 3 for night-shift nurses); the second series
of urine collections was performed 48 hours later, on
the last day off of the experimental period (day 7 for
day-shift workers and day 5 for night-shift workers).
To be certain that urine was collected at the appropriate times, subjects were given a preset automatic buzzer that was used during the wake periods. For the two
sleep periods during which urine was collected, the
subjects slept in the sleep laboratory, and urine was
obtained by the investigators at the appropriate times.
A maximum light level of 30 lux was maintained during urine collection in the laboratory.
Analysis
Actigraphy was analyzed using the commercially
available program Actisom@ (Axon-Phsio Com.,
France) and printouts were visually verified. Levels of
6-sulfatoxymelatonin in diluted urine (1140) were determined using radioimmunoassay (11). Acrophase
and amplitude of the 24-hour urinary metabolite profiles were determined using the cosinor method
(12,13). Paired t test and Mann-Whitney U test in nonnormally distributed variables were used for comparisons between groups. Two-way analysis of variance
with repeated measures on time were also performed.
The Systat@ statistical package was used for analyses
(14).
SHIFT IN SLEEP TIME WITH SHIFT WORK
TABLE 2.
541
Actigraphic results during the selected study period on total subject groups
Last work day without
sleep disruption
(mean:!: SD)
Night shift
Sleep onset time
0808 ill :!: 117 min
TST activity (minutes) 335:!: 84
Time of activity acrophase
1913 hr:!: 169 min
Amplitude of activity
at acrophase (units
of activity)
14.0 + 5.5
Last off day without
sleep disruption
(mean:!: SD)
Day shift
Significance
(Mann-Whitney
U test:
p value)
Night shift
Day shift
Significance
(MannWhitney
U test:
p value)
2238 hr :!: 83 min
431 :!: 73
0.0000
0.002
2323 ill :!: 86 min
572 :!: 116
2352 hr :!: 100 min
521 :!: 94
ns
ns
1336 ill:!: 66 min
0.0000
1558 ill :!: 108 min 1430 hr:!: 159 min
ns
20.5 + 7.5
0.005
2l.0 + 10.0
2l.0 :!: 10.0
ns
hr = hour(s); min = minute(s).
RESULTS
Sociodemographic and clinical information on the
subjects are presented in Table 1. There were no differences in "morningness" and "eveningness", anxiety, depression, or "global" scores between the two
groups. The standardized shift work indices were also
similar in both groups with the exception of the "involvement/no involvement" section. The night-shift
had a significantly lower score at 82 ::!:: 24 (versus 96
::!:: 14, Mann-Whitney U test: p = 0.04). As shown in
Table 1, day- and night-shifts had about the same
amount of sociofamilial responsibility and similar
mean commuting times. Night-shift nurses expressed
more concern than day-shift nurses about safety during
the commute (3:1 ratio).
There was no difference in the amount of alcohol
or cigarettes consumed per day, which was moderate
to low overall. Night-shift nurses were heavier than
day-shift nurses, with a body mass index (BMI) of 24
::!:: 2 kg/m2 versus 22 ::!:: 2 kg/m2 (p < 0.05), but none
were obese. One night-shift nurse had the habit of using zolpidem 10 mg before bedtime during her work
period. Seven night-shift nurses complained of frequent "generalized and nonspecific pain", compared
with two day-shift nurses.
Reasons for choosing the permanent night shift included:
ning of the actigraphy study period. Although there
was an increase in total sleep time in both groups on
days off, the increase was much greater in the nightshift workers. For night-shift workers, total sleep time
(TST) decreased significantly (p < 0.0001, Mann
Whitney U test) during their work periods compared
to days off (TST = 296 ::!:: 87 versus 413 ::!:: 48 minutes).
Sleep during actigraphy period
Results from the actigraphy, presented in Table 2,
do not include data from the two 24-hour periods when
subjects gave urine samples every 2 hours. Inactivity
(sleep time) is much shorter on work days in the nightshift group than in the day-shift group. The difference
in sleep time between workdays and off days was significant (ANOVA = 0.0001) within the night-shift
group. The day-shift group also slept less duri1ng workdays (p = 0.01).
Naps-day shift
On workdays, only one subject took a nap while at
work (21 minutes during the lunch hour). Two other
subjects took short naps after coming home from
work. On days off, six subjects took afternoon naps of
about 1 hour.
1. More time off (87%).
2. More time for familial and social interaction (56%).
3. More time for child care (47% of night-shift nurses
and 82% of the group with children).
4. Less strenuous and demanding work (10%).
5. A preference for night life and the feeling of being
a "night worker" (10%).
On workdays, nine subjects fell asleep while at
work, always between 0300 and 0600 hours. Mean
total sleep time was 114 ::!:: 45 minutes. On days off,
only four subjects had an afternoon nap, with a mean
total sleep time of 69 ::!:: 15 minutes.
Sleep prior to actigraphy study period
Correlation between sleep logs and actigraphy
Sleep logs did not demonstrate a statistical difference between the two groups in the overall estimated
total sleep time during the 15 days prior to the begin-
Correlation analysis comparing the results from actigraphy and individual sleep logs showed significant
correlations (see Table 3).
Naps-night shift
Sleep, Vol. 19, No.7, 1996
M. A. QUERA-SALVA ET AL.
542
TABLE 3. Pearson correlation coefficients: comparison
of actigraphic results and sleep diaries
Workday TST
Workday nap time
Off day TST
Off day nap time
r=
r=
r=
r=
0.93
0.81
0.82
0.90
6-Sulfatoxymelatonin results
Day shift
The time course of 6-sulfatoxymelatonin excretion
was fitted to a curve. The resulting sinusoidal curve
presented a consistent, significant amplitude during
workdays and days off in the day-shift nurses. 6-Sulfatoxymelatonin peaked near 0500 hours during the
night, independent of the work schedule; the calculated
mean acrophase occurred at 0440 hours ± 20 minutes
on workdays and at 04:56 hours ± 28 minutes on days
off.
TABLE 4.
Acrophase
(time)
Sleep, Vol. 19, No.7, 1996
Amplitude
(ng/hr)
Night-shift group
Workday
Subgroup A (n = 6)
Subgroup Ba (n = 13)
1208 hr ± 40 min
0636 hr ± 28 min
215 ± 38
390 ± 46
Off day
Subgroups A + B
0718 hr ± 24 min
357 ± 39
Workday
Off day
Day-shift group
0440 hr ± 20 min
358 ± 29
0456 hr ± 28 min
322 ± 38
hr = hour(s); min = minute(s).
Note: one subject had no peak, i.e. random distribution.
a
group A, with its rapid shift in 6-sulfatoxymelatonin
excretion, adapted quickly to the alternating need for
daytime and nighttime sleep associated with shift
work.
Night shift
On days off, the secretion of 6-sulfatoxymelatonin
presented a significant amplitude in all the subjects,
with mean acrophase at 0718 hours ± 24 minutes. On
work days, results of the total group analyses were not
statistically significant and the levels of 6-sulfatoxymelatonin in the urine appeared to be random during
the 24-hour period. However, analysis of each individual curve indicated that 19 out of the 20 subjects presented a significant amplitude. The last subject was
excluded from cosinor analysis. Two subgroups could
be identified within the night shift on workdays, based
on the timing of their peak 6-sulfatoxymelatonin excretion: Six out of 19 night-shift nurses (subgroup A)
showed a clear shift of 6-sulfatoxymelatonin excretion
in relation to their night-shift work, with the acrophase
at 1208 hours ± 40 minutes. The remaining 13 subjects (subgroup B) showed no evidence of such a shift.
Table 4 presents the mean acrophase time and mean
amplitude observed in each group and subgroup. Table
5 summarizes the 6-sulfatoxymelatonin excretion values obtained during the two 24-hour periods studied.
Subgroup A was smaller than subgroup B, was
moderately younger (32 ± 4 years versus 38 ± 8
years, p = 0.04) with a slight trend toward less years
of work (11 ± 5 versus 15 ± 10 years), and was composed solely of women (compared with nine women
and five men in subgroup B). Body mass index and
familial daytime demands were similar in both groups.
Actigraphic monitoring, however, indicated a significant difference in TST, with a mean increase of 104
minutes (Mann-Whitney U test, p = 0.02) in daytime
sleep during work periods in subgroup A (Fig. 1). Sub-
Calculation of acrophase and amplitude of
6-sulJatoxymelatonin
COMMENTS
Many night workers revert to a daytime schedule on
their days off, regardless of the consequences to their
health, cognitive function, and waking performance.
Internal and external desynchronization of circadian
rhythms is a major consequence of night work (15).
Persistent sleep disturbances, fatigue, mood and behavioral changes, and digestive ailments (from gastritis
to peptic ulcer) are all symptoms of intolerance to shift
work (16). In addition, workers who sleep during the
day tend to sleep several hours less than those who
sleep at night. Thus, shift workers are subjected to abnormal levels of sleepiness and are at an increased risk
for industrial and driving accidents (17,18).
Melatonin secretion over the 24-hour cycle is considered a phase marker for the endogenous component
TABLE s. 6-SulJatoxymelatonin measurements (ng/hour,
mean ± SD) in fast-shifting and nonshifting night nurses
Time
0800
1000
1200
1400
1600
1800
2000
2200
2400
0200
0400
0600
hr
=
hr
hr
hr
hr
hr
hr
hr
hr
hr
hr
hr
hr
hour(s).
Melatonin nonshifting
night workers
(n = 14)
1,060 ± 866
638 ± 548
41O±31O
190 ± 104
230 ± 206
266 ± 210
237 ± 204
162 ± 116
248 ± 126
503 ± 309
648 ± 365
1,022 ± 823
Melatonin shifting
night workers
(n =
541
506
659
587
360
206
223
161
156
213
167
225
±
±
±
±
±
±
±
±
±
±
±
±
6)
288
263
291
512
320
119
147
96
41
107
71
131
543
SHIFT IN SLEEP TIME WITH SHIFT WORK
of the circadian clock and an endogenous synchronizer
able to stabilize and reinforce the circadian rhythms
(19). The circadian rhythm of melatonin appears to be
rather stable in the face of exogenous influences, as
only strong illumination is known to alter melatonin
levels (20,21). In our study, light intensity on the hospital wards was found to be around 150 lux, but was
reduced during the two in-laboratory sleep periods
with urine collection to levels no greater than 30 lux.
Our study demonstrated that certain subjects possess the physiological ability to adapt to shifting
schedules. To our knowledge, this is the first time that
a rapid shift in melatonin secretion, corresponding to
an abrupt shift from a night-wake to a day-wake
schedule, has been shown to occur. Subgroup A nurses quickly adapted their melatonin secretion rhythms
to their new schedules, and were able to sleep as
much during the day as day-shift nurses slept at night
(see Fig. 1). This adaptation was in sharp contrast to
the findings in subgroup B nurses, whose urinary melatonin rhythm did not adapt to their fast-shifting
schedules but maintained peak secretion in relation to
their day-wake schedule. Night-shift nurses whose
melatonin secretion did not adapt to their changing
schedule had a significant sleep debt on workdays and
sleep rebound on off days. Although both day-shift
and night-shift nurses slept more on days off than on
workdays, the difference was far greater among night
workers. The sleep deprivation associated with the
night shift was responsible for inappropriate napping
on the job in nine nurses, and this napping is probably
only a partial account of the problem, because actigraphy does not evaluate the presence/absence of repetitive microsleeps.
The group that adapted quickly was moderately
younger than those who did not adapt, and was composed solely of women. Many questions, however,
remain unanswered. Is the fast adaptive response agerelated, showing a progressive decline with aging? Is
it related to some genetic factors known to control
the rhythm of the endogenous biological clock? Before drawing further conclusions, it will be important
to evaluate differences in daily performance between
adaptive and nonadaptive shift workers as well as to
assess whether work satisfaction and a feeling of
well-being are associated with this physiological ability to rapidly adjust melatonin levels.
Acknowledgement: We would like thank the Institut de
Recherches International Servier, who provided a research
grant in support of this study.
REFERENCES
l. Moore-Ede MC, Sulzman FM, Fuller CA. The clocks that time
us. Cambridge: Harvard University Press, 1982.
2. Folkard S, Arendt J, Clark M. Can melatonin improve shift
workers' tolerance of the night shift? Some preliminary findings. Chronobiollnt 1995;10(5):315-20.
3. Roden M, Koller M, Pirich K, Wierhapper H, Waldhauser E
Circadian melatonin and cortisol secretion pattern in permanent
night shift workers. Am J Physiol 1993;265:R261-7.
4. Sack RL, Blood ML, Lewy AJ. Melatonin rhythms in night shift
workers. Sleep 1992;15:434-4l.
5. Derogatis LR. "SCL-90". Administration scoring and procedure. Manual 1. Baltimore, MD: John Hopkins, 1977.
6. Horne JA, Osberg O. A self-assessment questionnaire to determine morningness eveningness in human circadian rhythms. Int
J ChronobioI1976;4:77-110.
7. Barton J, Folkard S, Smith LR, Spelten ER, Totterdel PA. Standard shift work index. MRCIESRC Social and Applied Psychology Unit. Sheffield, U.K: University of Sheffield.
8. Kripke DF, Mullaney DJ, Messin S, Wyborney VG. Wrist actigraphic measures of sleep and rhythms. Electroencephalogr
Clin Neurophysiol 1978;44:674-6.
9. Hauri PJ, Wisbey J. Wrist actigraphy in insomnia. Sleep
1992; 15:293-301.
10. Aubert-Tulkens G, Culee C, Harmant-Van Rijckevorsel K, Rodenstein D. Ambulatory evaluation of sleep disturbance and
therapeutic effects in sleep apnea syndrome by wrist actigraphy
monitoring. Am Rev Respir Dis 1987;136:851-6.
11. Harthe C, Claustrat B, Brunj-Chazot O. Direct radioimmunoassay of 6-sulfatoxymelatonin in plasma with use of an iodinated tracer. Clin Chem 1991 ;37:536-9.
12. Nelson W, Tong YL, Lee JK, Halberg E Methods for cosinor
rhythmometry. Chronobiologia 1979;6:305-23.
13. Sokal RR, Rohlf FJ. Biometry: the principles and practice of
statistics in biological research, 2nd edition. New York: Freeman, 1981.
14. Systat®. Systat intelligence software. Evanston, 1L, 1990.
15. Czeisler CA, Johnson MP, Duffy JF, Brouwn EN, Ronda JM,
Kronaver RE. Exposure to bright light and darkness to treat
physiologic maladaptation to night work. N Engl J Med 1990;
322:1253-9.
16. Reinberg Ap, Andlaver P, De Prins J, Malbecq W, Vieux N,
Bourdeleau P. Desynchronization of the oral temperature circadian rhythm and intolerance to shift work. Nature 1984;308:
272-4.
17. Akerstedt T. Sleepiness as a consequence of shift work. Sleep
1988; 11(1): 17-34.
18. Gold DR, Rogacz S, Bock N, Tosteson TD, Baum T, Speizer
FE, Czeisler CA. Rotating shift work, sleep, and accidents related to sleepiness in hospital nurses. Am J Public Health
1992;82: 10 11-4.
19. Armstrong SM. Melatonin: the internal zeitgeber of mammals.
Pineal Res Rev 1989;7: 157-202.
20. Lewy AJ, Sack R. The dim light melatonin onset as a marker
for circadian phase position. Chronobiol Int 1989;6:93-102.
2l. McIntyre 1M, Norman TR, Burrows GD, Armstrong SM. Human melatonin suppression by light is intensity dependent. J
Pineal Res 1989;6:149-56.
Sleep, Vol. 19, No.7, 1996