Psychological and psychophysiological effects of shift work

AKERSTEDT T. Psychological and psychophysiological effects of shift work. Scand J Work Environ Health 1990;16(suppl 1):67-73. The psychophysiology of shift work is mainlyrelated to circadian rhyth micityand sleep-wake phenomena. Individuals on a rotating three-shift or similar systemwork the night shift at the low phase of circadian rhythm. On retiring to bed in the morning they fall asleep rapidly but are prematurely awakened by their circadian rhythm and exhibit severesleepiness and reduced perform ance capacity. In connection with the morning shift the circadian psychophysiology makes it difficult to fallasleepasearlyas neededduring the precedingnight. Around 0400to 0500,whentheindividualsshould rise, they have difficulties awakening because of the sleep loss and the circadian rhythm, which at that point is at its lowest. Subsequently, day work is characterized by sleepiness and reduced performance. It should be emphasized that it does not seem possible to improve one's ability to adjust over time, even with permanent night work. Older age and " morningness" personalityare related to higher than average problems in adjusting.

Shift work is one of the more apparent and dramati c components of the wor k en vironment. It has been clearly linked to a seri es of acute and chronic effects on the organism, most of them related to the circadian rhythmicity of the body. The major effects concern sleep, alertness, and performance, but also long -term health. The purpose of the present paper is to provide a brief review of these effe cts and to discuss me chanisms and countermeasures.
Before turning to the effects, however, th e term " shift work " need s to be defined . The term usually refers to an arrangement of workhours which employs two or more teams (shifts) of workers in order to extend the hours of operation beyond that of con ventional office hours. It has , however, become customary to appl y the con cept also to groups with more un structured and irregul ar workhours and to groups with permanent night or evening work. With this usa ge the proportion of shift workers make up at least onefourth of the working population in mo st industrialized nations (1). In the present overview I have restricted the discussion to shi ft work that, at least occasionally, invol ves night work , since such schedules are the most interesting from a ps ychophysiological po int of view. Permanent night work, rotating threeshift work, nigh t-oriented ro ster wo rk, and irr egul ar work hours are included.
Reprint requests to: Dr T Akerstedt, National Institute for PsychosocialFactors and Health & Department for Stress Research, Karolinska Institute, Box60205, S-10401 Stockholm, Sweden.

Circadian physiology
The psychology and psychophysiology of shift work is intimately related to the rhythmic timing system of humans, particularly that having a 24-h periodthe circadian (from circa dies = approximately 24 h) system. It has its neural basis in the lower frontal hypothalamus , situated above the optic chiasma (2). Th ese suprachiasma tic nuclei produce a cyclic oscillation with a period of 24 h. Although the rhythm is rather stable, it may be modified by environmental synchronizers such as light, sleep, food, etc. The speed of adjustment to a new time zone is usualIy about 1 hid although this speed may differ between variables.
In order to describe the circadian rh ythm o f an individual, frequent measurements are needed -during work , leisure time, and sleep. Thi s need places a considerable burden on the subjects, and researchers have, for this reason, tended to focus on functions that are easy to measure, such as oral temperature and urinary constituents (2,3). Figure 1 derives from one of the most ext ensive studies of oral temperature , with a total of 133 workers in all (4). During the day of the first night shift an increase occurs from the time of rising in the morning to a peak in the evening. Thereafter, the temperature faIls during the night shift towards a minimum around 0400, after whi ch a rise is seen towards the end of the shift and the new morning bedtime. The fifth shift shows a similar pattern but with seemingly low temperature during the morning. (No measurements were taken during sleep. ) Such a pattern , with low night levels and high day levels, has been demonstrated for many physiological variables, eg, cortisol, potassium , adrenaline, etc (2,3).
In contrast to the variables just presented, which have a strong endogenous rhythmicity partly unaffected by behavior, other variables mainly reflect di-67 .  rect changes in the rest-activity pattern. This holds true for, among others, noradrenaline excretion, heart rate, and blood pressure. Incidentally, a rather peculiar observation in this context is that the subjective effort associated with a certain heart rate at a given work load is higher during the night shift than during the day shift (5,6). In some sense this phenomenon could be interpreted as the subjects being "older" on the night shift. Maximum work capacity does not differ though. A somewhat related observation is the occurrence of ventricular ectopic activity in connection with night work (7).
With respect to adjustment over several consecutive shifts figure I suggests that a few hours' delay of the nightly fall of oral temperature has occurred by the fifth night shift . Still, the minimum occurs at the same time as during the first night shift. If at all present, the adjustment over the five night shifts must be considered marginal. The same pattern has been observed in many otherstudies (3). It is likely, however, that part of the apparent adjustment is a direct effect of the environment, unrelated to the biological clock but "masking" its output (8). Lying down will, eg, reduce body temperature, and activity will raise it, both masking the underlying circadian pattern. Actually. it might be argued that the endogenous circadian rhythm never adjusts in shift workers (8). The reason for the marginal or nonexisting adjustment is that the circadian system, as discussed later, is very persistent and needs a longer time for adjustment than night workers ever enjoy since they usually revert to a diurnal life when off duty.
Laboratory studies allow a much better control of environmental influences and make it easier to carry out around-the-clock measurements . In one of the classic studies Colquhoun et al (9) showed that oral temperature across 12 consecutive night shifts flattened but never completely adjusted . Similar results have been published by, eg, Knauth et al (10) and Weitz-68 man & Kripke (I I). On the whole, most of the adjustment tends to occur during the first 1 to 3 d and then proceeds at a slower pace. It should be observed that in these studies all environmental synchronizers (light, food , social life) were geared towards a nocturnal life. This is something the night worker has little chance to experience.
It should be emphasized that most of the studies of the physiological circadian rhythms of shift workers are mainly of theoretical interest since a clear relation between rhythm adjustment and health parameters has seldom been demonstrated, except for a few studies suggesting that individuals who have difficulties tolerating shift work may have desynchronized rhythms or small amplitudes of their entrained rhythms (12).

Sleep
Disturbed sleep is perhaps the most dramatic effect of shift work . A number of survey studies have indicated that shift workers have difficulties ,mainly at maintaining sleep after the night shift and initiating sleep before the morning shift (13). The afternoon shift has usually presented no sleep problems.
The standard psychophysiological approach to sleep usually involves recording an electroencephalogram, an electrooculogram, and an electromyogram on paper and scoring the output visually in sleep stages per 30-s intervals (14). The standard sleep stages include wakefulness (stage 0), superficial to deep sleep (stages I to 4), and rapid eye movement sleep (stage REMdream sleep).
Sleep studies of shift workers have mostly been carried out in the laboratory (13). Recently, however , some stud ies of shift workers' sleep have been made in the workers' natural sleeping environment (15)(16)(17). The results are fairly conclusive in that sleep length on the night and morning shifts of rotating shift workers is reduced by I to 4 h. This reduction mainly affects stage 2 and REM. Stages 3 and 4 [which together make up slow wave sleep (SWS) or deep sleep] seem seldom to be affected. Furthermore, sleep latency is increased in connection with the morning shift and is shortened in connection with the night shift. Figure  2b demonstrates a hypnogram (sleep stages plotted against time) for the night shift. Note that the postworkday sleep is short but otherwise exhibits a normal pattern with two sleep cycles.
Rather little is known about the adjustment process across a series of night shifts. The available studies suggest that sleep length does not improve a great deal (18,19). Permanent night workers seem to sleep longer, however, than rotating shift workers on the night shift (19)(20)(21)(22)(23).
The reason for the shortened daytime sleep has in several studies been attributed to higher noise levels at that time (24,25). This may certainly be one of the causes of disturbed daytime sleep. On the other hand, sleep afte r the night shift is shortened also under optimallaboratory conditions (26, 27). Thu s noise does not seem to be the major cause of disturbed day sleep. A stronger influence is exerted by the circadian rhythm. Postponing sleep to different times of day under conditions of isolation from time-of-day cues (26) shows that the more sleep is postponed from the evening towards noon the next day , the more truncated it becomes, and when noon is reached the trend reverts. Thus sleep during the morn ing hours is strongly interfered with , despite the sizeable sleep loss that, logically, should enhance the ability to maintain sleep. Similar observations have been made for subj ects who can select their own pre ferred sleep-wake pattern under conditions of long-term isolation from time cues (27, 28). In the latt er stud ies it has been demonstrated that the factor most closely associated with the premature termin ation of sleep is the rising phase of the temperature cycle.

Sleepiness
Many questionnaire studies have demonstrated that shift workers report more fatigue than do day workers (29). Usually, the fatigue is particularly widespread on the night shift , hardl y appears at all on the afternoon shift , and is intermediate on the morning shift. In some studies sleepiness has been reported to be severe enou gh to have resulted in actual incidents of falling asleep during the night shift.
The upper part of figure 2 illustrates the 24-h pattern of rated sleepiness in a group of 24 three-shift workers at a paper mill (17). In connect ion with the afternoon shift sleepiness never reached high levels but was low during the day-evening and reached a medium level at bedtime. In conne ction with the night shift sleepiness increased during the night and reached a pronounced peak during the second half of the night shift. This pattern of early morning sleepiness has been demonstrated in many other studies (19,30).
Physiological evidence of night shift sleepiness is more scarce . However, in the study illustrated in figure 2, electroencephalography and electrooculography were also carried out. These procedures were done with the aid of small subject-worn tape recorders (Medilog) for a duration of 24 h on three occasion s involving morning, afternoon, and night shifts. The lower part of figure 2 shows the hypno gram of one worker during the night shift (17). During wor k two episodes of sleep can be seen. They are follo wed by a (short) day sleep of little more than 4 h, and later on by a 45-min nap during leisure time. Similar incident s of sleep occurred for approximately one-fourth of the subjects. Usually they occurred during the second half of the night shift and never in connection with an y other shift. Importantly, sleep on the job was not condoned by the company, nor was there any official awareness that sleep would or could occur during workhours. Similar results but with ultrashort intrusions of sleep (as judged by electroencephalography, electrooculography, and electromyograph y) have been demonstrated for locomotive engineers during work (31) and for other groups (32,33).
Incidentally, the general impression from most studies of sleepiness dur ing activity is that , although a certain sleepiness is clearly perceived by the individual, there seems to be no "final warning" before dozing off (29). This, very likely, constitutes a major sa fety problem in many occupations.
As to adjustment over shifts, there is a clear impression tha t night shift sleepiness will gradually delay its appearance over successive shift s (34-38) in a manner very similar to the behavior of oral temperature discussed earlier. There is no indication , however, that more than a marginal adjustment takes place. Thi s seems to be the case also for permanent night workers.
The cause of night shift sleepiness is apparently the combined influence of circadian and sleep-loss factors . The former was obvious in many of the field studies already cited and is practically always correlated with the bod y temperature rhythm . The influence of sleep loss is more difficult to isolate in field studies but may be readil y obser ved in laboratory sleep deprivation stud ies (39). In addition Carskadon & Dement (40) have demonstrated that 3 h of sleep reduction results in increased subj ective and physiological sleepiness (using the multiple sleep latency test). Furthermore this sleepiness measure showed accumulation across successive days of restriction.

Performance
If sleepiness on the night shift is as widespread and as dram atic as has already been indicated, one would expect to see pronounced effects on performance, and con sequently on output and safety. One of the classics in this area is the study by Bjern er et al (41), who showed that erro rs in meter readings over a period of 20 years in a gas work s had a pronounced peak on the night shift. There was also a secondary peak during the afternoon ( figure 3). Similarly, Browne (42) demonstrated that telephone oper ators connected calls a t a con siderably slower pace at night. Hildebrandt et al (43) found that -locomotive engineers failed to operate their alerting safety device more often at night than during the day, with a secondary peak around 1500.
Most other studies of performance have used laboratory types of tests and demonstrated, eg, reduced reaction time or poorer mental arithmetic on the night shift (15). Flight simulation studies have, furthermore, shown that the ability to "fly" a simulator at night ma y decrease to a level corresponding to that after moderate alcohol consumption (0.05 010 blood alcohol) (44). To these results may be added those from numerous laboratory studies which ha ve demonstrated that performance on many tasks deteriorates during the night hours (45).
Adjustment across shifts has very seldom been investigated under practical conditions . Laboratory investigations, however, clearly indicat e that adjustment does occur , although it may take up to two weeks. Frequently, the body temperature rhythm adjusts in parallel.
The impression of the night shift deterioration of performance is mainl y based on fa irly simple psychomotor types of tasks. There is, however, some speculation that high-level cognitive tasks, because of a high memory load, might show a differently phased rhythm (45). The latter would not, however, apply to the situation where sleepiness has come close to actual sleep, since any type of activity would then be interfered with.

Modifying factors
Several factors influence the adjustment to shift work . One such factor is the direction of rotation of the shift schedule. Since the free-running (spontaneous) period of the human sleep-wake cycle averag es 25 h and since it can be entrained by environmental time cues only within 1 to 2 h of the free-run period, phase delays are easier to accomplish than phase advance s (52). For the rotating shift worker this situation implies that schedules that delay, ie, rotate clockwise (morningafternoon-night) should be preferred to those that rotate counterclockwise. There has been , however, very few practical tests of this theory. Still , C~eisler et al (53) have demonstrated that a change from counterclockwise to clockwise rotation, together with a change from 7-d to 21-d rotation, improved production and well being for three-shift workers. Orth-Gomer (54) found that a change in the same direction among rapidly (l d) rotating police officers reduced blood pressure and improved well being.
The length of a work shift is another par ameter that one would expect to influence at least sleepiness and performance. In the laboratory it is usually the case that performance falls with time if learning effects are eliminated (55). Still, in one study of policemen, Peacock et al (56) found no effec ts of a change from 8 h Another important point is that common sense and available data suggest that the output from a production pro cess will not be affected by night work as long the major determinant of the production flow is machines rather thanpeople. Thus, it seems rather unlikely that sleepiness induced by the night shi ft would affect output in all occupations.
A more important area of impact ma y be safety. If sleepiness is severe enough, interaction with the environment will cease, and, if this interaction coincid es with a critical need for action, an accident may ensue. Such potential performance lapses due to nightwork sleepiness were seen for several of the locomotive engineers discussed earlier (31). The transport area is, in fact, where most of the available accident data on night shift sleepiness has been obtained. Thus Harris (46) and Hamelin (47,48) convincingly demonstrated that single vehicle accidents have, by far, the greatest probability of occurring at night (early morning) . Most of these accidents are thought to be due to sleepiness. With respect to air transport Ribak et al (49) found military flight accidents to be increased in the early morning, and Price & Holley (50) argued that also man y civil air transport accidents may be caused by fatigue du e to work scheduling. Finally, a number of spectacular nuclear accidents (including those at Chernob yl and Harrisburg) have been partly attributed to fat igue-inducing work schedules (51).
As with sleepiness, the main reason for night shift deterioration in performance is circadian rhythmi city and sleep loss (45).  o o (nine shifts across 8 d) to 12-h sh ifts (two night s-one free -two da ys-three free) on overall alertness. However , the distribution of free da ys changed at the same time. Two other studies of nu rses (57) and' industrial shift work er s (58) ha ve produced similar result s. Recently Rosa & Colligan (59) used 2-h ratings in a field experiment and demonstrated that the 12-h night shift , indeed, produced higher ratings of fatigue than 8-h night shifts. In addition, in a study of accidents of truck drivers, Hamelin (48) demonstrated a V-shaped relation between hours dri ven and accidents, ie, after an initial " warm up " period accident risk was low , with an increase to wards II h of dri ving.
As ma y be expected, also the watch-keeping systems (4 on, 8 off) on ships are asso ciated with low alertness and poor per formance during the night (60). Apparently, rotat ing systems cause greater disturban ce to the individual than do sta ble systems.
Another type of unusual workhours is th at of air crews on transmeridian routes. Then, not only are the work hours displaced to "biological" night time, but also the time reference is changed through time-zone shifts. As with other types of shift work, survey studies ha ve demonstr ated disturbed sleep and wakefulness (61). The di sturbed wakefulness has been evidenced also in flight simulator studies.
In some occupations the personnel ma y sleep at the worksite un til needed . Thi s is t he case for , a mo ng others, ph ysicians. Since the greater part of such nights are often are spent working, sleepiness-fatigue is often pronounced, and performance tends to be reduced, although the practical implic ations (for the patients) are still unknown (62,63). Other forms of "on-call" systems ma y be found among , for example, eng ineer officers in the me rchant marine (64) .
Among ind ividual differences age ha s been related to sleep disturbances (65,66). In electroencephalographic studies tr ends ha ve been found to wards more superficial sleep in middle-aged shift workers (16). Th e studies by Foret et al (65) a nd Ak erstedt & Torsvall (66) also indicated that experience was negatively related to general well bein g over a number of years. Koller et al (67) found that reduced health appeared earlier among shift workers than among day workers. Dahlgren (68) found no effects of three years of night work on the rhythm of rated activation acro ss night shifts. Neither did Wynn et al (69), over a temporary lO-week period of weekly alternation bet ween night and da y work in a gro up of nur ses. Dumont et al (70) found that the amount of sleep-wake and related disturbances in present da y workers was positi vely relat ed to the ir pr evious experience of night work. Guilleminault et al (71) found an overrepresentat ion of former shift workers with different clinical sleep/wake disturbances appearing at a sleep clinic. Although not directly related to sleepiness, it is still of interest to observe that Angersbach et al (72) ha ve demonstrated an earlier occurrence of gast rointest inal disease among three-shift workers than among da y workers. Similarly, Knu tsson et al (73) demonstrated that the incidence of myocardial infarction (and cardiovascular disease in gener al) is related to the amount of expo sure to shift work.
Finally, the trait o f morningness (having a tendency toward s ear ly sleep-wake preferences) has frequently been associated with poor adjustment to shift work (74,75). Th is has also been the case fo r the trait of sleep rigidity (76).

Concluding comments
Taken together, the reviewed literature clearly indica tes that shift work that involves night shifts strongly influences th e psychology and psychophysiology of the ind ividual.