Shift work: coping with the biological clock

Abstract

The internal circadian clock adapts slowly, if at all, to rapid transitions between different shift schedules. This leads to misalignment (desynchrony) of rhythmic physiological systems, such as sleep, alertness, performance, metabolism and the hormones melatonin and cortisol, with the imposed work–rest schedule. Consequences include sleep deprivation and poor performance. Clock gene variants may influence tolerance of sleep deprivation. Shift work is associated with an increased risk of major disease (heart disease and cancer) and this may also, at least in part, be attributed to frequent circadian desynchrony. Abnormal metabolism has been invoked as a contributory factor to the increased risk of heart disease. There is recent evidence for an increased risk of certain cancers, with hypothesized causal roles of light at night, melatonin suppression and circadian desynchrony. Various strategies exist for coping with circadian desynchrony and for hastening circadian realignment (if desired). The most important factor in manipulating the circadian system is exposure to and/or avoidance of bright light at specific times of the ‘biological night’.

Body clock, cancer, circadian rhythm, heart disease, light, melatonin, metabolism, shift work

Introduction

Many reviews have been published regarding the subjective perceptions, health, performance and psychosocial aspects of shift work [1–11]. There is little doubt that shift work is associated with a number of health problems, for example poor sleep, gastrointestinal disorders, abnormal metabolic responses and increased risk of accidents. A longer term risk of major disease such as heart disease and cancer is beginning to be appreciated. This review will concentrate on shift work in relation to biological rhythms since disturbed rhythms appear to underlie many of the short- and long-term health problems of shift workers [12–14]. To this end, an introduction to the subject is provided.

Literature search

A literature search with the keywords ‘shift work’ and ‘circadian’ gave 1034 references in PubMed. Since the pineal hormone melatonin is currently used as the primary output marker of the internal clock (as well as its actions as a chronobiotic), the search was then restricted to ((shift work) and (circadian) and (melatonin)). This provided 189 references, which together with the author's personal collection formed the basis of this review.

Importance of biological rhythms to health

Biological rhythms serve to align our physiological functions with the environment. We are a diurnal species and thus, we normally sleep at night and are active during the daytime. The timing of functions with prominent rhythms such as sleep, sleepiness, metabolism, alertness and performance in a normal environment is such that they are optimal during the most suitable phase of the day (Figure 1). Abrupt deviations from ‘normal’ timing of work and sleep can lead to problems, for example sleep taken during the day is usually shorter and of worse quality than when taken at night [6,15]. Alertness and performance reach their nadir at night during peak sleep propensity and fatigue [13,16,17] and close to the low point of core body temperature and the peak of melatonin secretion. The health problems and increased risk of major disease in long-term shift workers are ascribed largely to working out of phase with the internal biological clock. It is likely that many perceptions of the detrimental effects of clock disruption or abnormal timing derive from observations in shift workers.

Figure 1.

Diagrammatic examples of circadian rhythms, from Rajaratnam and Arendt, Lancet 2001 [13], by permission.

Characteristics of circadian rhythms

Basic properties

Everything is rhythmic unless proved otherwise [18]. Biological rhythms of various periodicity are ubiquitous. The frequency displayed varies from fractions of a second (for example the firing of neurones) to years (for example population variations). By far, the most information is available concerning daily rhythms [18,19]. They are either externally imposed, internally generated or more frequently a combination of these two factors. Internally generated rhythms with approximately a 24 h period are known as circadian, from the Latin ‘circa diem’, ‘about a day’). Circadian rhythms serve to temporally programme the daily sequence of metabolic and behavioural changes. By definition, they persist in the absence of time cues such as alternating light and darkness and are coordinated by an internal biological clock (pacemaker, oscillator) situated in the suprachiasmatic nuclei (SCN) of the brain hypothalamus [20]. The basis of circadian rhythm generation is a negative feedback loop of clock gene expression [21,22].

Individuals kept in a time-free environment (or at least with very weak time cues) manifest their own endogenous periodicity referred to as ‘free-running’. The free-running period is individually variable and is an inherited characteristic. On average, the human endogenous period (or tau) is about 24.2–3 h although this does depend on previous experience of time cues [18,23]. Synchronization or entrainment of the circadian clock to 24 h is dependent on suitable time cues, also known as ‘zeitgebers’. In circadian literature, synchronization means that rhythms display a 24 h period but may not necessarily be in the right phase, for example, abnormally delayed or advanced. Entrainment means that the rhythms are synchronized with the appropriate phase. When entrained to the 24 h day, a short endogenous tau is associated with morning diurnal preference (larks) and a long tau with evening preference (owls) [18].

Circadian response to time cues

Because the circadian clock period is not exactly 24 h, it must be reset regularly (phase shifted) to maintain a 24 h period. The most important time cue for maintaining a 24 h period is the light dark cycle acting partly via a novel retinal photoreceptor system and a novel photopigment melanopsin (circadian photoreception) [24]. Recent evidence indicates that short wavelengths of light (460–480 nm, blue) have the most powerful resetting effects [25]. Blind people with no conscious or unconscious light perception frequently display free-running rhythms, underlining the importance of light. The timing of sleep also has an influence together with minor ‘non-photic’ zeitgebers such as exercise, social cues, clock time and food ingestion. Specific manipulation of food timing in animals influences a so-called food entrainable oscillator, which is independent of the SCN [26]. The content of meals in humans may also have a minor influence.

The circadian response (change in timing or phase shift) to light exposure, and indeed to other time cues, is dependent on the strength and timing of the stimulus. It can be described by a ‘phase response’ curve (Figure 2) [27,28]. The central clock adapts slowly, and with considerable individual variability, to a rapid shift in work time or time zone. After a time zone change, the average rate often approximates to 1 h of adaptive shift per day. After an abrupt shift in work time, the change is very variable as discussed later [29–31]. During the process of adaptation, endogenous rhythms are out of phase with the external environment (external desynchronization). They may also be out phase with each other, i.e. assume a transitory abnormal phase relationship (internal desynchronization). This condition is often referred to as ‘circadian desynchrony’. Time cues or zeitgebers are all important in controlling the circadian response to such changes. In general, it is easier to delay the clock than to advance it in view of the >24 h period of most people. During a period of desynchrony, for example, a single night of night shift in a sequence of days, workers are attempting to sleep at a time of maximum alertness and to work at the nadir of alertness and performance. If adaptation of the clock to a new work schedule occurs, the problems of desynchrony resolve [32–35].

Figure 2.

Circadian response (‘phase response curve’, shift of the melatonin rhythm, advances are positive, delays are negative) to a 1–2 h light pulse, ca 300 lux, 500 nm, at different times of night. DLMO = dim light melatonin onset, on average at ∼2100 h, thus 8 h after DLMO = 0500 h clock time. From Paul et al. [28], by permission.

Genetic basis of circadian rhythms

Many of the genes concerned with circadian rhythm generation in mammals and other species have now been identified, e.g. CLOCK, PER1, PER2, PER3, TIM, CRY1, CRY2, BMAL1, REV-ERBALPHA. The mechanism is similar in all species investigated and substantial homology exists between, for example, Drosophila and mammals. Oscillation of clock genes also occurs in peripheral structures, and in general, they are considered to be coordinated through SCN activity. However, it is possible to shift the timing of some peripheral oscillations (for example, in the liver by timed feeding), independently of the SCN [36]. Investigation of polymorphisms in human clock genes in relation to occupational health and disease is in its infancy. Some polymorphisms have been identified and associations are emerging with phenotypic characteristics such as diurnal preference (larks–owls), intrinsic period, vulnerability to disease and response to sleep deprivation [37–39].

The circadian clock influences hormones, behaviour, cognitive function, metabolism, cell proliferation, apoptosis and responses to genotoxic stress [22]. There is new strong evidence concerning the importance of circadian control for health in that disruption of circadian clock gene expression can lead to increased incidence or progression of cancer (in animals) [22,40].

Examples of rhythms relevant to human disease

Some examples of human rhythms in disease processes include night time asthma, early morning increases in blood pressure, death rate from cardiovascular disease and stroke, disrupted menstrual cycles, abnormal cortisol rhythm in Cushing's syndrome, sleep disorders for example delayed sleep phase syndrome, advanced sleep phase syndrome, non-24 h sleep wake cycles (especially in the blind), some psychiatric disorders. Numerous aspects of human biochemistry show rhythmicity, even urinary creatinine. Thus diagnostic tests should be aware of these rhythms. Measurement of a given rhythmic variable in someone who has just crossed several time zones, or worked a series of night shifts, can give false-negative or false-positive results. Moreover, many drugs have a rhythmic variation in both pharmacokinetics and efficacy (chronopharmacology).

The melatonin rhythm

A darkness hormone

Melatonin (N-acetyl-5-methoxytryptamine) in an indolic hormone whose principal physiological function is to provide a humoral time cue for the organization of seasonal and circadian rhythms [41]. The pineal gland secretes melatonin with a marked circadian rhythm, peaking at night—it has been called the ‘darkness hormone’ and the duration of its secretion is directly related to the length of the night. In animals which depend on day length to time their seasonal physiology, the length of melatonin secretion signals the length of the night. In humans, its circulating concentrations are high from ∼2100 to 0700 h with large individual variations. This period can be used to define ‘biological night’. The peak secretion occurs ∼0400 h, closely associated with the nadir of core body temperature, alertness and performance (Figure 1). In specific circumstances, humans may also show changes in the duration of secretion [41].

Melatonin as a chronobiotic

Melatonin is not only a so-called ‘hand of the clock’, it has the ability to induce sleepiness or sleep, change circadian phase and to entrain free-running rhythms when administered in suitable doses and timing [42]. There are several phase response curves to melatonin with slight differences, which can be used to predict the chosen timing in order to hasten a circadian phase shift. This is important in a shift work context given that a number of attempts have been made to treat shift workers with melatonin with variable results.

Melatonin suppression by light

Light of sufficient intensity and spectral composition will suppress melatonin production at night [41,43]. Suppression is detectable at 30–50 lux and maximum from around 1000–2000 lux. Natural daylight can attain >100 000 lux. This suppression is associated with rapidly increased alertness and core body temperature, although causal relationships are not clear. It is important in a shift work context as light suppression of melatonin has been hypothesized to be detrimental to health [44]. A night shift worker whose circadian clock is in day mode, or unadapted, will secrete melatonin during work hours. Similarly, a worker who has adapted their clock to night shift will secrete melatonin during the day and on return to day shift or rest days will secrete melatonin during the hours of natural daylight. Actual personal light exposure during night shift work has rarely been measured, examples of personal light exposure on North Sea oil rigs are shown in [45]. In general, the amount of suppression reported during field studies on night shift is minor, ∼20% [46,47].

Melatonin indicates the timing of the biological clock

Shifts in the timing of melatonin are considered to represent changes in timing of the central clock. Measurement of melatonin in plasma, saliva or its urinary metabolite 6-sulphatoxymelatonin [43] provides the best peripheral measure of central clock timing (Figure 3). Other marker rhythms such as core body temperature and cortisol are more subject to so-called masking, whereby an internal or external influence distorts the rhythm. For example, exercise and food strongly influence core temperature and stress modifies the cortisol rhythm. The most reliable results regarding circadian status in shift work have been obtained with melatonin measures and this review will concentrate on melatonin-derived information.

Figure 3.

Characteristics of the melatonin rhythm used to define timing of the internal clock. From [43] by permission.

Circadian desynchrony in shift work

Relationship to work hours

There are many varieties of shift work and a legal definition does not appear to exist except for ‘working outside normal working hours’. For the purposes of this review, let us consider that it is working during ‘average’ biological night, i.e. 2100–0700 h. Also for the purposes of this review, the assumption is made, based on controlled laboratory experiments, that when sleep is taken during the period of peak melatonin secretion (and thus, in theory, the nadir of alertness, performance and core body temperature), it is optimized. In night shift conditions, if peak melatonin secretion is shifted to occur during day sleep, it is presumed that adaptation to night shift has occurred.

The most numerically important shift work conditions, at least in the UK, are irregular night shifts (sometimes nights and sometimes days) and rotating schedules (information from the Office of National Statistics) (Box 1).

Box 1.

Types of shift work

UK Labour Force Surveys, Office for National Statistics, April to June, 2005

All shift workers, 3 083 270, 11% of working population (28 301 000)

Rotating shifts, 1 260 845

Permanent nightshifts, 290 661

Sometimes nights, sometimes days, 416 412

Weekend shifts, 49 407

Other types of shift work, 1 065 945

Examples of rotations include 3 days early shift (e.g. 0600–1400 h), 3 days late shift (e.g. 1400–2200 h), 3 days night shift (e.g. 2200–0600 h) and rest days. These common schedules do not allow the internal clock to adapt fully to night shift since there is substantial inertia in the circadian system. After abrupt large changes in time cues, the daily shifts in circadian timing rarely exceed 1–1.5 h on average without interventions. Exposure to morning light (in the travel home window after night shift) is at a time that opposes a delay shift of the clock to adapt (Figure 2), and daytime social cues counter circadian adaptation. Data from field studies indicate that the greater the morning light exposure the less circadian adaptation is seen [48]. In temperate latitudes natural bright light will be more prevalent in summer in the early morning, and any shift of the circadian system during night shift will in theory be countered more strongly than in winter.

Partial shifts in circadian timing can be seen in short-term night shift work, and a relationship to the timing of light exposure is present, either delays or advances, in relation to the light phase response curve [29,49,50]. Natural light exposure, the most powerful influence, evidently depends on the shift timing. Also important is the diurnal preference of the subjects [29]. Evening preference people are reported, as might be expected, to have a greater tendency to delay and morning people to advance [29,47].

Most permanent or long-term night shift workers (with exceptions, see below) do not adapt their circadian system to the imposed work schedule. A recent meta-analysis of 6-sulphatoxymelatonin rhythms in permanent night workers indicates that only a small percentage (<3%) shows complete circadian adaptation and <25% adjust to the point that some benefit would be derived from the adaptive shift [51]. This may depend on the ability to maintain night activity and day sleep on days off as well as other factors such as diurnal preference. However, there is a paucity of data and further research is needed.

Thus, the vast majority of shift workers will be working during their circadian nadir and trying to sleep during periods of maximum alertness. The curtailment of sleep when taken during the day in shift workers is well documented and is a cause of sleep deprivation. Some examples of short sleep are shown in Table 1.

Table 1.

Examples of shorter sleep in night shift workers

4.8 h, night shift, fast rotation, Axelsson et al. [52]

5–6 h, morning shift starting before 6 am, Kecklund and Akerstedt [53]

6.04 h, 7 nights, 7 days, 12 h on 12 h off, Gibbs et al. [45]

5.83 h, 4 h on 8 h off, permanent night watch, ships crew, Arendt et al. [54]

Compare

7 h (approximately) healthy adults, Groeger et al. [55]

8.7 h healthy young men, sleep ad lib, Rajaratnam et al. [56]

Sleep deprivation

Sleep deprivation concomitant (inter alia) upon circadian desynchrony has been attributed a causal role in obesity, metabolic syndrome, glucose intolerance/diabetes, increased accidents and errors. Sleep restriction has also been associated with alterations of neuroendocrine control of appetite [57–60].

In relation to sleep deprivation, an important study of interns weekly work hours in the USA found that they made 36% more serious medical errors during a traditional work schedule than during an intervention schedule that eliminated extended work shifts. These included 21% more serious medication errors and 5.6 times as many serious diagnostic errors [60].

Changing from day to night shift often implies a period of 20–24 h without sleep. The decrements in performance during the latter part of this sleep deprivation may be equivalent to an illegal level of alcohol in the blood [16]. Accidents following a combination of sleep deprivation and working during the circadian nadir in performance and the maximum sleep propensity have led to litigation against individuals and employers [13]. It is considered that fatigue may be a more important cause of transport accidents than alcohol (Figure 4).

Figure 4.

Comparison of the effect of blood alcohol concentration (BAC) and hours of wakefulness on task performance. Higher scores indicate better performance. The decrement in mean performance at blood alcohol concentrations of 0.10% or greater is similar in magnitude to those observed after 25–27 h of wakefulness. From Rajaratnam and Arendt [13] after Dawson and Reid [16] by permission. (A) Effect of blood alcohol on task performance. The dotted horizontal line is the mean performance at a blood alcohol concentration of 0.08% (the legal limit for driving in the UK). (B) Performance is likely to be affected by circadian factors and sleep debt, thereby accounting for the recovery in performance after 27 h of wakefulness (about 1100 h in this experiment).

Metabolism

The circadian system regulates metabolism [61] and increasing evidence relates circadian desynchrony to disorders such as metabolic syndrome (insulin resistance, high blood pressure, central obesity, decreased high density lipoprotein (HDL) cholesterol, elevated triglycerides (triacylglycerol, TAG) and cardiovascular disease [59]. For example, eating a standard meal at night (biological night) leads to high blood lipid (TAG) and evidence for insulin resistance, compared to the same meal taken during the day [62,63]. Interestingly, there is some evidence that men are more susceptible to these metabolic abnormalities than women. TAG is an independent risk factor for development of heart disease. This may provide at least a partial explanation for the increased risk of heart disease in shift workers. In large surveys, shift workers have higher TAG levels as well as higher total cholesterol than the general population [11,64].

There are associations between polymorphisms in the gene CLOCK, obesity and the metabolic syndrome in man, and mice bearing a particular mutation of the gene CLOCK develop metabolic syndrome and obesity [65,66].

Circadian adaptation in unusual environments

Some exceptions to the general rule that shift workers do not fully adapt to night shift are found in isolated environments. On the British Antarctic Base of Halley, 75°S, each base member does a week of night shift (2000–0800 h, fire watch) at a time, in rotation with other personnel. The vast majority of people shift their circadian system, assessed by aMT6s rhythms in urine, by up to 10–12 h to align with the new work schedule, within a week [67,68]. Thus, the peak of melatonin production occurs within the daytime sleep period—an important condition for sleep duration, latency and quality. This is thought to be due to the lack of social and family obligations, no requirement to return home in natural light and in winter when the sun does not rise for 3 months, a lack of conflicting light exposure. In these circumstances, it is apparent that owls adapt by delay faster than larks [69]. Problems occur when endeavouring to adapt back to day work particularly in winter. Realignment of the circadian system can take weeks and some people will free-run for a time.

These observations in Antarctica prompted studies in somewhat similar circumstances on North Sea oil rigs. Work schedules vary but tours of duty usually last for 2–3 weeks in a socially isolated environment at high latitudes. Here, working 1800–0600 h for ≥1 weeks also leads to full circadian adaptation in the majority of cases with the accompanying problems returning to day life [30,70,71]. Interestingly, a 1900–0700 h schedule is less conducive to circadian adaptation, possibly due to early morning light exposure after work especially in summer [72]. Sleep is worse during this shift than the 1800–0600 h shift. The so-called ‘swing shifts’ worked in the North Sea present a confused picture. Seven night shifts (1800–0600 h) followed by 7 day shifts (0600–1800 h) leads in most people to adaptation to nights but with a very mixed response to the following days [30,73] (Figure 5). Some people delay, some advance and many show little readaptation to days during the first week. The response is partly predictable from the initial circadian phase position: delayed, intermediate or advanced. Other schedules such as a 7 day, 1200–2400 h day shift followed by a 7 day 2400–1200 h night shift show partial or no adaptation to night shift which is dependent on season [71]. Again, this can be attributed to light exposure countering adaptation.

Figure 5.

Progression of the individual timing of the melatonin rhythm (by urinary aMT6s), during a week of nights (1800–0600 h) followed by a week of days (0600–1800 h) in 11 individuals working on a North Sea oil rig. Most shift their timing such that at the end of nights the peak of melatonin is during the daytime sleep and thus, they are adapted. The subsequent response to a change to day work is highly variable. From [30] by permission.

Such circumstances, when workers show different circadian timings according to whether or not they have adapted to nights, have allowed field assessments of the metabolic consequences of a night shift meal. As with controlled laboratory experiments, elevated TAG, low density lipoprotein (LDL) cholesterol and evidence of insulin resistance were found when unadapted [45]. These sequelae resolved when adaptation had occurred.

Marine watchkeeping

Special cases of shift working are seen in marine watchkeeping systems. There are very few data relating to melatonin rhythms; however, a study in submarines has shown evidence that crew can free run while working an 18 h day, submerged for long periods [74]. Circadian adaptation was studied in crew working 4 h on and 8 h off on fixed or rotating schedules on a British Antarctic Survey ship travelling from the UK to 75°S. They showed evidence of partial circadian adaptation to the 1200–1600 h, 2400–0400 h fixed watch but not to weekly rotating watches [54].

Light at night and the risk of major disease

It is not the purpose of this review to evaluate the epidemiological evidence for increased risk of disease in shift work; this can be found elsewhere. It is generally accepted that there is an increased risk of heart disease and some contributory factors related to the biological clock have been discussed above.

A possible significant association was identified between female breast cancer and shift work some time ago [75]. This is potentially a major problem since estimates from the Spring 2002 wave of the Labour Force survey suggest that an estimated 1.8 million women in Great Britain usually or sometimes do shift work. Of these, an estimated 400 000 are involved in night work of various schedules [76]. In 2006, the World Health Organization (WHO) (International Agency for Research on Cancer, IARC) published a brief report in the Lancet of an expert meeting on whether or not shift work was associated with an increased risk of cancer, particularly breast cancer. The conclusion was that shift work was a probable carcinogen [77]. A full monograph is expected from this meeting but has not yet been published. There are considerable implications arising from this decision and, for example, according to the UK national press in March 2009, the Danish government decided to compensate shift workers who develop breast cancer. Both the UK and Dutch governments also issued publications discussing the evidence and in general decided that further epidemiological evidence and mechanistic data were needed [78,79]. Assuming that the risk assessments are correct, let us consider the possible mechanisms.

The most well-known theory concerning cancer and shift work relates to light exposure at night (light at night, LAN). Stevens and Davis [44] hypothesized that the increasing incidence of breast cancer in the developed world was due to light exposure at night. He further proposed that, since light suppresses melatonin and melatonin has some oncostatic activity in animals, the increase in breast cancer was due to a decrease in melatonin. Numerous questions arise from this proposal, some of which can be addressed. The WHO (IARC) expert meeting concluded that there was definite evidence for anti-cancer effects of melatonin in animals and in vitro but little in humans. It should be noted that human in vivo data are sparse. The data which directly addressed the cause and effect relationship between melatonin and human breast cancer involved maintaining human breast transplants (xenografts) in rats and assessing short-term markers of cancer with and without circulating endogenous levels of melatonin. Physiological levels of melatonin were able to reduce or abolish carcinogenic changes in these markers [80]. This latter study does suggest that endogenous melatonin has anti-proliferative effects working via a membrane receptor. However, not all studies have shown anti-proliferative effects of melatonin in vitro. Is melatonin suppressed in night shift workers? Again the data are sparse, the existing evidence mentioned previously suggests that an ∼20% reduction may be found during the night shift. Is a 20% reduction in melatonin carcinogenic? There is no answer to this question. However, it should be noted that adrenergic beta-receptor blocking drugs such as atenolol and propanolol suppress melatonin and are not known to be carcinogenic [41]. Moreover, the individual variability in melatonin production is very large indeed and in cross-sectional studies, large numbers of subjects are needed to show this small overall reduction [41].

More convincing are the effects of general circadian disruption in animals [81]. Exposure of animals to continuous light increases vulnerability to cancer development [82,83]. Subjecting rodents to forced phase shifts analogous to rotating shift work or frequent time zone change substantially increases proliferation of implanted cancers [40]. Manipulation of clock gene function likewise has carcinogenic effects [14,22,84,85] and the circadian clock is considered to be a tumour suppressor. So the case is close to being made for circadian disruption leading to cancer.

The problem is that increasing light at night in short-term shift work leads to improved alertness, performance and possibly metabolism, while no doubt increasing melatonin suppression. At present, investigations are proceeding on the use of glasses, which can block the short wavelengths most likely to suppress melatonin while hopefully maintaining alertness and performance.

Approaches to reducing desynchrony

The question arises as to the benefits and disadvantages of aiding or countering adaptation in order to secure the maximum duration of good quality sleep and other health benefits. In the unusual case of full adaptation, it makes sense to hasten this process and that of readaptation to day work. In short-term night shift, it may be more useful to maintain daytime circadian phase while using strategies such as alerting stimulants (caffeine and possibly modafinil), quiet dark sleeping quarters (and possibly hypnotics) to preserve sleep and performance.

Light of suitable spectral composition and intensity can be used to adjust the timing and probably amplitude of circadian rhythms. The hormone melatonin can also act as a zeitgeber (vide supra). Light treatment during the first half of biological night prior to the melatonin peak will delay circadian rhythms and during the latter half, after the melatonin peak, will advance rhythms (Figure 2). Melatonin treatment by contrast advances rhythms in the first half of biological night and delays them in the latter half.

There is no doubt that in controlled laboratory situations, and with good compliance at home, both light and melatonin, separately or in combination, can be used with correct timing to hasten phase shift of the circadian system to align it with the new work–rest schedule [32,43,50,86]. There are clear benefits for sleep, alertness and performance. Moreover with suitable timing, the sleep-inducing effects of melatonin during ‘biological day’ can also be exploited. In field situations, the results are inconsistent. Very probably this is due to the large individual differences in response to phase shift and in consequence mistiming of the treatment. However, in studies offshore and in Antarctica, useful results have been reported with timed light treatment [67,87]. The author is only aware of one combined treatment study, which was offshore and with beneficial effects [88]. It is possible that the most useful application of melatonin to shift workers would be to facilitate sleep or a nap, prior to night shift. It is particularly difficult to sleep in the early evening, and the combination of low-dose melatonin, a dark room and recumbency is very effective at enabling sleep at this time of day [56].

An alternative approach has been described recently. This is to shift the circadian system, using timed light and melatonin, just to the point where the melatonin peak falls within the sleep period, avoiding large shifts which lead to readaptation problems [34,89]. This strategy appears to provide benefit for sleep, alertness and performance. However, if the timing is wrong, the opposite of the desired result will be produced. For example, instead of adapting to an 8 h advance in work time by advancing the clock, the system may delay. Avoidance of light at the wrong time is possibly more important than the light treatment itself.

A genetic variant predicting intolerance to sleep deprivation

Morningness has been related to intolerance to shift work in some studies although not all are consistent. Recently, a length polymorphism in the clock gene PER3 (PER3 5/5, 4/5, 4/4) was found to relate to diurnal preference, 5/5 being associated with extreme morningness and 4/4 with extreme eveningness [38]. Subsequently, it was found that the 5/5 genotype suffered more from sleep deprivation than the 4/4 genotype and these differences could be explained by an effect of the polymorphism on sleep homeostasis [37]: it was associated with greater sleep propensity and a higher proportion of slow-wave sleep than the 4/4 genotype. Most importantly, the subjects with the 5/5 variant suffered far greater consequences of sleep deprivation in terms of performance (notably during the circadian nadir) than the 4/4 genotypes. Mongrain et al. [90] have also shown that morning types have a higher homeostatic response to sleep disruption than evening types. This observation suggests that workers who are extreme morning types should choose their schedules carefully with regard to preserving sleep. One further study addressing a similar question, but different methodology, found an association with sleep homeostasis but did not confirm the effect on neurobehavioural responses [91].

It is of interest to note there are also preliminary reports of greater susceptibility to breast cancer in women with the 5/5 variant and prostate cancer in men with the CRY2-variant C allele [39,92]. No doubt more information will be available shortly given the importance of predicting the possible health consequences of shift work.

Conflicts of interest

None declared.

I would like to thank all colleagues, students, volunteers and funding bodies who have provided input, samples and support for our work on biological rhythms over the years. I would particularly like to mention Mark Midwinter, Jonathan Ross, Jon Lund, Lil Ng, Gavin Francis, Victoria Mottram (Antarctic Base doctors), Richard Barnes, Michelle Gibbs and Helen Thorne (University of Surrey PhD students) who persuaded numerous workers at Halley Base (75°S) and on the North Sea oil installations to provide sequential urine samples during their period offshore. These heroic efforts have provided a dynamic picture of the long-term circadian response to phase shift.