Circadian rhythms: principles and measurement.

The physiological basis of rhythmic processes is explained, and the influence of rhythmicity on pathological processes and pharmacological responses is outlined. Methods of data collection and analysis are described, and the importance of taking account of, and even exploiting, rhythmicity in experimental design is discussed.

Complex interaction of the sleep-wake cycle and circadian phase modulates mood in healthy subjects.

BACKGROUND: Several studies of healthy volunteers have revealed that subjective mood may vary with the duration of prior wakefulness and with the time of day. However, in these studies, the effects of extended wakefulness and circadian phase remained confounded, and the interaction of these 2 processes could not be assessed quantitatively. METHODS: In the present study, a total of 24 healthy young subjects (16 men, 8 women) lived on a 30-hour sleep-wake schedule for 19 to 23 days or on a 28-hour sleep-wake schedule for 33 to 36 days; both schedules induced desynchrony between the subjects' sleep-wake cycle and their endogenous circadian pacemaker. Subjective mood was assessed by 2 types of visual analog scales, which were administered twice every 2 hours and every 20 minutes, respectively, during all scheduled wakefulness episodes. A circadian phase and an interval elapsed since awakening were attributed to each data point, and circadian and wake-dependent fluctuations of mood were assessed. RESULTS: A significant variation of mood with circadian phase was observed, but no reliable main effect of the duration of prior wakefulness was found. A statistically significant interaction of circadian and wake-dependent fluctuations was evident; when the analysis was restricted to specific circadian phases, mood improved, deteriorated, or remained stable with the duration of prior wakefulness. CONCLUSIONS: These results indicate that, in healthy young subjects, subjective mood is influenced by a complex and nonadditive interaction of circadian phase and duration of prior wakefulness. The nature of this interaction is such that moderate changes in the timing of the sleep-wake cycle may have profound effects on subsequent mood.

Deriving a "phase response curve" from adjustment to simulated time zone transitions.

Thirty-eight subjects in groups of two to four have been subjected to simulated time zone transitions of 8 hr eastward, 8 hr westward, or 12 hr; all experiments have taken place in a temporal isolation unit, in which the light intensity during the waking periods was 250-300 lux. Circadian rhythms of rectal temperature have been used as a marker of the process of adjustment, the data being analyzed before (raw) and after "purification." Conventional results have been obtained, but when the data have been treated appropriately, they have produced a relationship between shift of the sleep-activity cycle and shift of the temperature rhythms that shows many characteristics of a phase response curve. Even though the factor or factors causing such adjustment are unknown, the results confirm that changes of a consonant set of "weak" zeitgebers is sufficient for entrainment to a new time zone to occur.

Effects upon circadian rhythmicity of an alteration to the sleep-wake cycle: problems of assessment resulting from measurement in the presence of sleep and analysis in terms of a single shifted component.

Experiments were performed upon groups of three or four human subjects in an isolation chamber (total n = 14). Subjects lived initially on a conventional lifestyle and then delayed their hours of sleep by 8 hr (so mimicking some aspects of nightwork) for 2 or 5 days. They also performed two constant routines--protocols designed to minimize any effects due to the environment, mealtimes, and activity. Regular samples of urine were taken when subjects were awake, and were analyzed for sodium, potassium, and chloride; rectal temperature was measured and logged at 6-min intervals throughout. Shifts in circadian rhythms produced by the change in sleep time were assessed by cosinor and cross-correlation techniques. The protocol enabled these assessments to be made on days when sleep was allowed and under constant-routine conditions, so that masking and behavioral effects could be investigated also. The results confirmed that adjustment to the change in sleep time was slow and only partial, and that assessments made on days when sleep was allowed overestimated this adjustment. Furthermore, it was concluded that, whereas cosinor and cross-correlation techniques using only one shifting component were equally useful in describing the observed changes, both were inferior to a cross-correlation technique that made use of two shifting components. Some practical and theoretical implications of these results are discussed.

"Demasking" the temperature rhythm after simulated time zone transitions.

Simulated time zone transitions were performed in an isolation unit upon groups of one to four human subjects. In the first series of experiments, the adjustment of the circadian rhythm of body temperature, measured in the presence of sleep and other masking factors, was assessed by cosinor analysis and by cross-correlation methods. These methods modeled the circadian timing system either as a single component or as the sum of two components, those due to exogenous and endogenous influences. The one-component models described a more rapid adjustment of the temperature rhythm to the time zone transition than did the two-component models; we attribute this difference to the masking effects of the exogenous component. In a second series of experiments, we showed that the shift of the endogenous component, as assessed by the two-component models, was not significantly different from that measured during constant routines. The results also showed that, if the zeitgebers were phased in advance of the endogenous component, then advances of the endogenous component were produced only if this mismatch was less than about 10 hr. Mismatches greater than this, and cases where the zeitgebers were delayed with respect to the endogenous component, both produced delays of the endogenous component. We conclude that the two-component cross-correlation methods can be used to estimate shifts of the endogenous component of a circadian rhythm in the presence of masking factors. They are therefore an alternative to constant routines when these latter are impracticable to carry out.

Mathematical and statistical analysis of circadian rhythms.

The mathematical and statistical analysis of biological time series is complex and often involves specialised techniques. In this article we review several of these techniques, placing particular emphasis on the usefulness, assumptions and kind of data that they require. Because classical methods of time series analysis often require long spans of data that are not frequently available in biological studies, particularly in clinical circumstances, several alternative techniques are described. Where possible, emphasis is placed upon simple rather than esoteric mathematical descriptions of data. Also covered are problems of interpretation that might arise as a result of the mathematical analysis.

A human phase-response curve to light.

Using 'classical' experimental protocols, a human phase-response curve (PRC) to a single 3-h bright light pulse has been established. When the light pulse was centred slightly before the time of body temperature minimum, the circadian system delayed, whilst a pulse slightly after the minimum advanced it. Maximum phase shifts were about 2 h. When light pulses over 3 successive cycles were used, larger shifts (4-7 h) were produced. It is concluded that the human PRC does not differ in principle from that found in other species, except with respect to the light intensity required.

Shiftwork, helplessness and depression.

Forty-three student nurses undertaking shiftwork for the first time participated in a study which tracked their perceptions of altered neurovegetative function, perceived criticism from others, sense of purpose and control and psychosomatic complaints. It was found that this first ever episode of shiftwork produced marked changes in all of the above. These findings have implications for circadian rhythm hypotheses of depression as well as for the methodology of future studies on cognitive or psychosocial variables in depression.

Some circulatory responses to exercise at different times of day.

Circadian rhythms in heart rate were examined at rest, immediately pre-exercise, during submaximal and maximal exercise on a cycle ergometer, and during recovery post-exercise (N = 10). Observations were made under controlled conditions at 0300, 0900, 1500, and 2100 hours. A significant circadian rhythm was found for resting heart rate lying supine and sitting pre-exercise (P less than 0.05), peak values being measured at 1500 hours. The acrophase in the oral temperature rhythm at 1739 hours was not significantly out of phase with that of resting heart rate (P greater than 0.05). The rhythm in heart rate persisted during submaximal exercise (150 W) and at the maximal rate (P less than 0.05); the amplitude of the rhythm was attenuated at maximum. Ratings of perceived exertion at submaximal and maximal exercise intensities, and time of day (P greater than 0.05). The increment of 0.2 degrees C in oral temperature during exercise did not exhibit circadian variation (P greater than 0.05). A significant rhythm was found for recovery heart rates in minutes 2, 3, 4, and 5 post-exercise (P less than 0.05). Observations of systolic and diastolic pressures pre- and post-exercise were inconclusive. Therefore, the circadian rhythm in heart rate responses to exercise should be considered when a heart rate variable is used as a criterion in fitness testing or as an index of physiological strain.

Investigating the endogenous component of human circadian rhythms: a review of some simple alternatives to constant routines.

Several types of constant routine are accepted as an important means by which the endogenous component of circadian rhythms can be studied. Nevertheless, they are impracticable to perform and unsuitable for routine use in many individuals. We describe a group of simple methods with which rhythms measured in normal circumstances can be dissociated into the components due to masking and the internal clock. Each method is best suited to a particular type of experimental condition. Results from a variety of protocols are analysed by these and conventional methods to assess the validity of the new methods.

The use of constant routines in unmasking the endogenous component of human circadian rhythms.

A major problem in the study of the internal clock(s) that drives human circadian rhythms is that due to the effect produced by rhythmicity of habits and external influences ('masking'). A particularly potent factor in this respect is the sleep-wake cycle. It is anomalous that, even though this masking influence is widely accepted, most studies of circadian rhythmicity have been performed in the presence of such interferences. A protocol is described, the constant routine, by which these exogenous influences can be minimized, thereby enabling a closer scrutiny of the internal clock(s) to be made. An account is given of the different circumstances in which the constant routines have been used together with the results derived from such studies. Briefly, they indicate that nychthemeral studies can give misleading information about the rate of adjustment of the internal clock to various manipulations, e.g. time-zone transition, shift work. In addition, future studies making use of constant routines are described, in particular those which might enable the presence of more than one internal clock to be established.

Removing masking factors from urinary rhythm data in humans.

We have previously developed simple models that enable the exogenous and endogenous components of the circadian rhythm of body temperature to be separated. The present paper extends the method to urinary data. First, we have shown that the basic superiority of the two-component model over the one-component model persists when temperature data are converted into a format that is appropriate for urine sampling (that is, a single overnight sample and two-hourly samples during waking). Second, we provide normative endogenous data for urinary sodium, potassium and urate, data obtained from about 80 constant routines. These data are required for the two-component model. Third, we have compared the rate of adjustment to a simulated eastward time-zone transition of 8 hr in 8 subjects. This showed that the rate of adjustment assessed by the two-component model was significantly less than that assessed by the one-component model and much closer to that assessed in separate experiments (n = 15 subjects) using constant routines. We conclude that the two-component model can be used upon urinary data to give a closer approximation to the shift of the endogenous component, as assessed by constant routines, than can estimates that do not take into account the problem of masking caused by exogenous factors.

Masking in humans: the problem and some attempts to solve it.

Different types of masking are discussed together with an account of the masking effect that the sleep-wake cycle exerts upon the circadian rhythms of body temperature and urinary excretion. The relative importance to masking of the several components of differences between sleeping and wakefulness are then assessed. Means to deal with the problem of masking fall into two major categories. These attempt to minimise masking effects by protocols such as constant routines or control days, and mathematical models which separate results obtained in the presence of masking influences into endogenous and exogenous components. (The problem of the extent to which masking influences can render the endogenous component of a rhythm an impure reflection of the internal oscillator is considered also.) These different techniques are compared with respect to their usefulness and assumptions. Finally, a brief speculation is given of the usefulness of masking.

Can insulated skin temperature act as a substitute for rectal temperature when studying circadian rhythms?

We measured rectal, lateral chest wall, and axillary temperature every half hour for at least 24 h while subjects were living normal life-styles and keeping a sleep/activity diary. We then used a purification method to estimate the decrease of temperature due to sleep and the increases due to sitting, standing, walking, or exercising, as well as the parameters of the cosine curve that described the "purified data." Cosinor analysis of raw and purified data showed that the acrophases from both skin sites were much more variable and up to 8 h later than were those from the rectum (particularly if exercise had been taken), even though the acrophases from the two skin sites were similar to each other. For rectal temperature, there was an increase in the size of the masking effect as activity progressed through the sequence: sitting, standing or walking, exercising. In contrast, for both chest wall and axillary temperatures, although sitting produced masking effects similar to those for rectal temperature, masking effects due to standing or walking and exercising were much smaller, and sometimes they were even less than the masking effects due to sitting. These results indicate that our measurements of cutaneous temperature did not act as a substitute for rectal temperature, particularly when the subject was physically active rather than sedentary.

The adjustment of the circadian rhythm of body temperature to simulated time-zone transitions: a comparison of the effect of using raw versus unmasked data.

Deep body temperature and sleep/activity diaries data were recorded during control days and for 6 days after simulated time zone transitions of 8 h to the east (six subjects) or west (seven subjects). Circadian rhythms were assessed by cosinor analysis of both raw data (the conventional method) and purified data (corrected for the effects of sleep and activity). Analysis of raw data gives misleading information about the phase and amplitude of the rhythms due to the masking effects of the exogenous component. Use of purified data indicates that during the process of adjustment after an eastward shift (a) phase changes are more erratic than after a shift to the west; (b) no marked decrease in the amplitude of the rhythms is evident; and (c) no clear evidence exists that the circadian rhythm breaks up temporarily. The masking effect was less after the time zone transition if sleep maintenance was poor.

The shape of the endogenous circadian rhythm of rectal temperature in humans.

Fourteen healthy subjects have been studied in an isolation unit while living on a 30h "day" (20h awake, 10h asleep) for 14 (solar) days but while aware of real time. Waking activities were sedentary and included reading, watching television, and so forth. Throughout, regular recordings of rectal temperature were made, and in a subgroup of 6 subjects, activity was measured by a wrist accelerometer. Temperature data have been subjected to cosinor analysis after "purification," a method that enables the endogenous (clock-drive) and exogenous (activity-driven) components of the circadian rhythm to be assessed. Moreover, the protocol enables effects due to the circadian rhythm and time-since-waking to be separated. Results showed that the masking effects on body temperature exerted by the exogenous factors appeared to be less than average in the hours before and just after the peak of the endogenous temperature rhythm. This has the effect of producing a temperature plateau rather than a peak during the daytime. The implications of this for mental performance and sleep initiation are discussed.

Circadian rhythms and masking: an overview.

Masking, as is well known, enables an organism to act immediately and in an appropriate way to changes of the environment, integrating with internally produced rhythmicity. It now appears that masking can be used to cover a far wider range of problems than was originally intended. To separate masking effects from the effects due to an internal oscillator, several techniques have been used. Such protocols, however, like the constant routine protocol, often replace one form of masking by another. The situation becomes even more complex when one realizes that the output of an internal oscillator modifies the input. The question might be asked whether it is possible to study the properties of the internal oscillator in vivo at all. This article attempts to produce a framework for future discussions.

Variation in meals and sleep-activity patterns in aged subjects; its relevance to circadian rhythm studies.

The study was performed upon a sample of aged and non-institutionalized subjects. Information was obtained by questionnaires and diaries on personal factors during a typical week. A random subset was subjected to a more detailed analysis of the composition of their meals. Results showed that increasing age was correlated with: a decreased day-by-day variability in an individual's time of retiring, rising and eating meals; earlier sleep times; increased frequency of daytime naps and nocturnal awakenings; and decreased physical activity. These results occurred both in subjects living alone and in those living with company. Day-by-day differences in the composition of meals tended to decrease with age. When differences between individuals were considered then these tended to increase with age. Some implications of these findings for studies of circadian rhythmicity in aged subjects--in whom the timing of circadian rhythms becomes more erratic and amplitude falls--are discussed.

The development of circadian rhythmicity in neonates.

Eleven healthy, full-term babies were studied on the second day (d2) after birth and again 4 weeks (w4) later. The babies lived in natural lighting conditions and were fed every 4 h. Blood pressure, heart rate, skin (abdomen) and rectal temperatures were measured at 10-min intervals for 24 h. Behavioural states (deep sleep, light sleep, drowsy, alert, crying) were measured at the same times, and used to purify the raw data. Both the raw and purified data were assessed for circadian (24-h) rhythmicity by cosinor analysis. Circadian rhythms in heart rate and blood pressure were poorly developed at d2 and w4. By contrast, skin and rectal temperatures showed circadian rhythmicity that increased in amplitude between d2 and w4; this increase was seen in both raw and purified data. The masking effect due to sleep changed also; the depression caused by "deep sleep" became greater between d2 and w4. The results indicate that the development of circadian rhythmicity in body temperature was not secondary to that of a circadian pattern of sleep and activity, and presumably derives from a body clock that is beginning to develop independently.

Effect of bright light on tolerance to night work.

Fifteen young (mean age 23.4 years) female nurses engaged in a resuscitation unit and working on a fast rotating shift schedule comprising two consecutive night shifts were exposed to short periods (4 x 20 min) of bright light (2350 lx) during their night duty to test a possible positive effect on their tolerance to night work. Two nights with normal lighting (20-380 lx) and two nights with bright light were compared. The following positive effects of bright light upon psychophysical conditions and performance efficiency were noted: in particular, signs of better physical fitness; less tiredness and sleepiness; a more balanced sleep pattern; and higher performance efficiency (letter cancellation test). This result could not be attributed to shifts of the internal clock although the exact cause remains to be determined. In fact, hormonal excretion and body temperature did not show any effect from bright light. In addition melatonin excretion was not suppressed appreciably by the bright light used.