Circadian adjustment of men and women to night work.

OBJECTIVES
The aims of the present study were to define the adjustment of sleep-wakefulness and other circadian rhythms during consecutive night shifts and to study the effect of gender on the adjustment to night work.


METHODS
Twenty experienced shift workers, 9 men and 11 women, were studied under controlled laboratory conditions. After a night of habituation, the subjects worked one day shift and three night shifts. Rectal temperature was measured continuously during the study. Salivary melatonin and cortisol were assessed at 2-h intervals, and subjective sleepiness was checked every hour during the work shifts. Sleep was registered by the static-charge sensitive bed (SCSB) method after the workshifts.


RESULTS
Subjective sleepiness varied both between the work shifts and between the men and women. The subjects felt most alert during the day shift and were the sleepiest during the first night shift. Sleepiness decreased during the consecutive night shifts but did not reach the level of the day shift during the three nights. The sleepiness of the women decreased more than that of the men, the difference being significant during the second and third night shifts (P < 0.05). The circadian rhythm of body temperature, salivary cortisol, and time in bed changed significantly (P < 0.05) between the workshifts, but no differences were found between the men and women in the circadian adjustment of the physiological variables to night work.


CONCLUSIONS
The men felt sleepier than the women during the consecutive night shifts, although no physiological differences were found between the genders. However, since normal social and domestic factors were excluded, the findings cannot be considered as evidence of women's better adaptation to night work in everyday life.

Most nightshift workers have problems sleeping during the day (1) and during night shifts they experience severe sleepiness (2). The basic mechailism causing the sleep disturbances of night workers seems to be circadian. On the other hand, the sleep of shift workers is also greatly disturbed by environmental and domestic factors under natural conditions (3). Most studies on the effects of night work on sleep and circadian rhythms have been conducted under field conditions with many environmental factors disturbing sleep (4,2). Experimental studies on the circadian adjustment process during consecutive night shifts are rare (5-10). The experimental studies are also contradictory as regards sleepiness; it has been shown to decrease (1 1-13), to increase (14, lo), or to remain unchanged (15) during consecutive night shifts.
There are also great individual differences in the tolerance to shift work (16). Especially women have tradi-tionally been protected from night work. Surprisingly, we are not aware of any controlled experimental studies on the effects of gender on the circadian adjustment to night work. The circadian patterns of men and women are allnost identical (17), but some indications of differences between the two in circadian rhythms (18,19) and sleep structure (20, 21) have been detected. However, most of the data indicating differences in their responses to shift work are based on psychosocial studies (22)(23)(24). In shift work, women have been shown to have more psychosomatic complaints and higher absenteeism from work than men. Furthermore, the prevalence of sleep disturbances and drowsiness at work has been reported to be higher among female shiftworkers (25,23).
The present study was designed to define the exact adjustment of sleep-wakefulness and the circadian rhythms of body temperature, melatonin, and cortisol during consecutive night shifts. In addition, the objective was to conlpare the adjustment to night work between Inen and women under closely controlled laboratory conditions.

Subjects
Twenty experienced postal workers, 9 Inen (mean age 40.6 years) and 11 women (mean age 38.8 years), were studied (table 1). The main work task of the subjects was manual letter sorting. Six men and nine woinen viere three-shift worl<ers, while three rnen and two women had two-shift work without night shirts at the time of the study. The shift worlt experience of the men was still somewhat longer than that of the women (table I). There were no differences in the diurnal type scale (26, morn-i~lgness-eveningness) between the subjects (table 1). According to a routine medical examination, all the subjects were healthy, and none were on continuous medication. Four men and three women said that they "snore loudly," but none had sleep apnea according to the measurements of the static-charge sensitive bed (SCSB), which included a continuous registration of respiration. The subjects were paid for their participation.

Design
Every subject stayed in the shiftworlt laboratory for one week. After a night of habituation, the subjects worked one day shift and three consecutive night shifts. The shift work laboratory was equipped with facilities for controlled letter-sorting work, sound-proof sleeping chambers, and a separate room for leisure time. The windows were darkened from sunlight to 1500 during the day time, and the light intensity was always below 1000 Ix in the laboratory.
During the day shift, the subjects worked between 0900 and 1600 and were allowed to have a 60-~nin lunch hour and a 30-min coffee brealt. After the worlt they had leisure time and were allowed to sleep between 2230 and 0630. During the night shift, the subjects worked from 2300 to 0600 and were allowed to take one 30-mi11 brealt. After the night shift they slept between 0630 and 1430 and had leisure time before the next night shift. The work task of the subjects was letter sorting, mimicking Table 1. Characteristics of the subjects. that at their actual workplace; the worlt was self-paced. The subjects spent their leisure time watching television, reading, doing needlework, and such. The meal times during the weelt were kept regular, with moderate consumption of coffee. Napping, sauna, exercise, or alcohol was not allowed.

Circadian rhythms
Rectal temperature (T,,) was measured continuously with recordings every third minute during the weelt with a thermistor probe (YSI 401, Yellow Spring Instruments, United States) that was inserted about 10 cm into the rectum. The T,, was recorded with a desk-top coinputer (HP9122, Hewlett Packard, United States) via an analogue-digital (AID) converter (HP3421A). To smooth the fluctuation of the T,,, the mean temperature values for every 30 min were calculated from the raw data and the lime of the T,, minimurn was estimated visually from the smoothed data.
Melatonin and cortisol were determined from samples of saliva, collected at 12-11 intervals during the worltshifts. Unstimulated saliva (2.5 ml) was collected into polyetl~ylene tubes and stored in a freezer (-20°C). The melatonin concentration was determined after chloroform extraction using a modification (27) of radioimmunoassay, originally described by Miles et a1 (28). Cortisol was determined by direct radioiminunoassay, as described earlier (27).

Sleep-wakefulness
The time in bed and the autonomic sleep phases were registered using the SCSB method (29). From the SCSB, the ballistocardiogram, respiratory movements, and inovements specific for postural changes can be registered siinultaneously by selective filtering of the original signal. After AID conversion, the frequencies of different movements were analyzed (BRI I , Biorec, 30), and the sleep was classified into autonomic sleep phases (active, intermediate, and quiet sleep) in epochs of 3 min. The time of falling asleep was estimated according to the appearance of the first phase of quiet sleep for 3 min.
Subjective sleepiness was studied every hour during the worltshifts with the ICarolinslta sleepiness scale (ICSS) (31), which is a modification of the Stanford sleepiness scale. The Karolinslta scale includes the following phases: extremely alert (score = l), alert (score = 3), neither alert nor sleepy (score = 5), sleepy but no difficulty in reinaining awalte (score = 7), extrelnely sleepy, fighting sleep (score = 9). The effect of gender, day (day shift and three night shifts), time of day (0900-1500 and 2400-0600), and their interactio~ls were studied by repeated-n~easures ailalyses of variance. The assuniptioil foi-probabilities provided by ordinary F-tests to be correct was tested by a sphericity test (32). If the sphericity test was significant, an adjusted F-test (33) was used. In cases in which the P-value of the sphericity test was lower than 0.0001, a multivariate analysis of variance was used. Pairwise comparisons were done when the overall F-ratio was significant.

Results
Results of the analysis of variance of all the studied variables are preseilted in table 2. The T,, miniinu~n (figure I ) varied between the days, reaching its minimum between 0400 and 0500 after the day shift and between 0700 and 0900 after the night shifts. The delay of the meail (N = 20) T,, minimurn averaged 3.3 (SD 1.8) h dming the first night shift and a f~lrther 0.5 (SD 2.0) h during the following night shifts. Accordi~ig to the analysis of variance with repeated measures, the interaction of gender and day was not significant (table 2). However, there was a mean difference of 1.2 (SD 2.1) h in the T,., ininilnum between the inen and women after the third night shift, the men showing slightly more (not significant) adjustment than the wolneil.
Salivary nielato~lin (figure 2) secretion diminished during the first hours of the day shift, but increased towards the early morning hours of the night shifts, the time effect being significant (table 2). There was no day or gender effect nor any significant interactions with salivary ~nelaioni~l secretion.
The salivary cortisol (figure 3) secretion levels were altered between the shifts and showed a significant interaction between day and the time of day (table 2). The mean cortisol level was the highest ill the beginning of the day shift, decreasi~ig towards the end of the shifi. During the consecutive night shifts the cortisol showed an illcreasing trend but a lowering level. There was no difference in cortisol secretioli between the men and the women.
According to the SCSB recordings (figure 4), there were only small nonsignificant differences between the shifts in the time of falling asleep and in the length of autonomic sleep phases, but the time in bed varied clemly between the days ( The subjective sleepiness (ICSS, figure 5) varied between the worltshifts (effect of day, table 2) with significant differeilces (t-test, P < 0.05) between all the shifts.
The subjects felt the most alert during the day shift and were the sleepiest during the first night shift. The meall sleepiiless decreased during the co~lsecutive night shifts, but it did not reach the level of the day shift during the   2). Subjective sleepiness also differed between the men and women in various shifts (interaction effect of gender and day, table 2). During the day shift there was no difference in perceived sleepiness between the men and the women, but in the course of the night shifts the women felt less sleepy than the men, especially during the second and the third night shift, the difference being significant (t-test, P < 0.05).

Discussion
The men felt sleepier than the women during consecutive night shifts, although no physiological differences were found in the circadian rhythms of body temperature, salivary melatonin, and cortisol or in sleep between the genders. In fact, there was a nonsignificant trend for men adjusting faster than women. This finding indicates that the women's increasing alertness during night shifts can-not be due to differences in circadian adjustment. Our results are in accordance with those of the study of Akerstedt & Froberg (17) showing that the circadian rhythms of the two genders were almost identical, and the only difference was that the men were characterized by increasing trends in ratings of fatigue and stress and the women were not during a 72-h sleep deprivation experiment. These results thus indicate that there is no clear physiological difference between men and women which would indicate that woinen adjust to night work worse than men. However, the difference in sleepiness cannot be considered evidence of women's better adjustment to night work in everyday life, because normal social and domestic factors were excluded from the laboratory study. Although there appear to be no great sex-related differences in the adaptation to night work, women seem to tolerate night work less than men because of the gender-related extra work load due to household duties and child care (22). In fact, the usually greater family responsibilities of women (22, 23) probably cancel out the potential benefit of circadian adjustment. The question of whether shift work should be organized by rapidly or slowly rotating night shifts is still one of the major issues in shiftwork research (34). The main argument favoring slowly rotating shift work is the claim that sleepiness decreases during consecutive night shifts because of the circadian adjustment process contrary to, for example, single night shifts in very rapid rotating systems. In our study, subjective sleepiness decreased gradually from the first night shift to the third but did not reach the earlier level of the day shift. The recovery of sleepiness during the consecutive night shifts shows that there was no cumulative sleep deficit due to night work. W e want to point out, however, that this was a laboratory study with a good sleeping environment, and the same kind of recovery can be achieved under field conditions only if subjects can sleep sufficiently. The present results, which show improvement in alertness during consecutive night shifts, still support the hypothesis that slowly rotating shift schedules can be better than rapidly rotating schedules when good sleeping conditions are available.
The reason for the increased level of sleepiness during the first night shift seems to be homeostatic. At the end of the first night shift the subjects had been awake for 24 h, while in later night shifts the preceding awake period was only 16 h. The considerable reduction of sleepiness during the consecutive night shifts was presumably also due to homeostatic rather than circadian mechanisms, since the phase delay of the T,, minimum was minimal at the same time (0.5 h per day). On the average, however, the subjects did not perceive severe sleepiness during the night shifts, and sleepiness varied within a rather normal range, in other words "sleepy but no difficulty in remaining awake" (31). In spite of being monotonous, letter sorting is self-paced and apparently motional enough to prevent dozing off at work.
Sleep length and latency were shortened after the night shifts, as expected (35,36). An uilexpected finding was the shortening of the quiet sleep phases in connection with day sleep. In the validation study of the SCSB method (37) we found that the phases of quiet sleep consist of mostly delta sleep (69%). In earlier encephalographic studies, slow wave sleep (stage 34) increased (35) or did not change (36) after the night shifts due to the sleep deprivation.
The timing of the T, minimum did not change significantly during the consecutive night shifts, but the deviation of the T,, minimum increased, being greatest after the third night shift. The results suggest that the phase delay of body temperature rhythm was only partial during consecutive night shifts, as suggested earlier (4). The delay of the T,, minimum was the most pronounced when the subjects changed froin the day to the night shift, probably due to "masking" effects (38, 39). The adjustment from the day to the first night shift was thus "overestimated" (38). It was, however, possible to study the impact of gender on the circadian rhythm of T,, since the masking factors of sleep, activity, and work were similar (same starting time) during the consecutive night shifts. Furthermore, the T,, minimum always occurred after sleep onset, reducing potential variation due to the different masking effects of sleep.
An interesting finding was that the mean salivary melatonin secretion profiles did not change during the consecutive night shifts, mainly because of the great interindividual alterations. Our results thus support some earlier findings that night shift workers have great variability in the timing of melatonin secretion (40) and may not actually show remarkable changes in melatonin rhythm during night work (41, 42). The secretion patterns of salivary cortisol, unlike the melatonin profiles, still changed, a phase delay thus being suggested for the consecutive night shifts.
In conclusion, the adjustment of circadian rhythms in the course of the night shifts was slow and only partial. Subjective sleepiness still decreased in the course of the night shifts, and the men felt sleepier than the women during the consecutive night shifts, although no physiological explanation was found.