SEARCH

SEARCH BY CITATION

Keywords:

  • alcohol;
  • daily;
  • early awakening;
  • karolinska sleepiness scale;
  • mixed model;
  • self-rated health;
  • sleep quality;
  • stress;
  • subjective health;
  • total sleep time;
  • weekend

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure Statement
  9. References

Sleepiness is linked to accidents and reduced performance, and is usually attributed to short/poor prior sleep and sleepiness. However, while the link between reduced sleep and subsequent sleepiness is well established in laboratory experiments of sleep reduction, very little is known about the day-to-day variation of sleepiness in everyday life and its relation to the immediately preceding sleep episode. The purpose of the present study was to investigate the characteristics of this relation across 42 consecutive days. Fifty volunteers participated. Self-reports of sleep were given in the morning and recorded with actigraphy; health was rated in the evening; and sleepiness was rated at eight points during the day (on a scale of 1–9). Results from mixed-model regression analyses showed that, on average, total sleep time predicted sleepiness during the rest of the day across the 42 days, with sleepiness increasing with shorter preceding sleep (β = −0.15 units h−1, < 0.001). Sleepiness also increased with earlier time of rising and lower-rated sleep quality. Days off reduced sleepiness, but was accounted for by sleep. Self-rated health improved when sleepiness was low during the same day (β = −0.36 unit unit−1 of rated health, < 0.001), but the two were measured simultaneously. Napping was related to high sleepiness during the same day. Actigraphy measures of sleep duration showed similar, but somewhat weaker, effects than diary measures. It was concluded that the main determinants of daytime sleepiness in a real-life day-to-day context were short sleep, poor sleep and early rising, and that days with high sleepiness ended with ratings of poorer health.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure Statement
  9. References

Sleepiness is a major cause of accidents (Philip and Akerstedt, 2006), and the neurocognitive consequences of sleep loss are profound (Goel et al., 2009). The major regulators of sleepiness are sleep duration, circadian phase and time since sleep (Dijk et al., 1992; Jewett et al., 1999). Sleepiness has been defined as the ‘the drive to fall asleep’ (Dement and Carskadon, 1982), and its momentary level may be measured at the physiological, behavioural or subjective levels. The subjective rating is the measure most easily available and very widespread. It has a close relationship to physiological indicators [electroencephalography (EEG) alpha and theta activity or slow eye movements] (Åkerstedt and Gillberg, 1990), as well as to performance (Ingre et al., 2006). It has also been used in a number of field studies and has shown high levels during, for example, night work (Landrigan et al., 2004), during late night driving on motorways (Sagaspe et al., 2008), in pupils who have a job outside the school (Teixeira et al., 2010) and in patients with long-term exposure to stress (Söderström et al., 2006), among others.

Considering the high levels of sleepiness in certain work situations and the associated accident risk, what the relation between sleep and sleepiness looks like in daily life is an interesting question, with its presumably modest variations of sleepiness and sleep between days. Is there any discernible variability between days? If it is, is it linked to the duration of the preceding sleep episode, which seems the logical assumption, considering experimental studies of acute (Härmä et al., 1998) sleep restriction, or of repeated partial sleep deprivation (Van Dongen et al., 2003)? However, both studies suggest that the sleep obtained needs to be considerably shorter than 6 h for immediate measurable effects on sleepiness. Furthermore, early rising times (which reduce sleep and extend the time awake) may contribute (Ingre et al., 2008), as may perceived sleep quality, such as that seen after suppression of Stages 3 and 4 sleep (Dijk et al., 2010). Another factor that may affect sleepiness through its effect on sleep duration or delayed awakening is weekend sleep extension; napping is also a factor that should be important in relation to sleepiness (Horne et al., 2008).

The studies cited above either represent experiments of somewhat dramatic sleep manipulation or naturally occurring acute shifts of work schedules. However, it is not known if daytime sleepiness is affected by the relatively modest day-to-day variability of sleep duration in normal life (Mezick et al., 2009) and what shape this relation may have. This question would require a study of the relation between daily measures of sleep, sleepiness and other variables over a longer time-period. The results of such a study would have considerable ecological validity, and may help in understanding the variation between ‘good’ and ‘bad’ days in life, although no published reports are available. The longitudinal covariation approach has been used, previously for investigating the effect of the daily variation of sleep on the daily variation of mood across 2 weeks (McCrae et al., 2008). Such a longitudinal approach requires multi-level modelling to deal with the longitudinal covariation involved.

The aim of the present study was to investigate how sleepiness during the day would be related to sleep during the previous night (reported in the morning or measured with actigraphy) over a period of 42 days. Also, bedtime, time of awakening, stress before sleep and workday/day off (in most cases workday/weekend) were used as predictors of sleepiness.

Apart from sleep-related variables, subjective health was also included as a predictor because sleepiness and sleep duration are increased during illness (Krueger, 2008). This sickness-induced sleepiness is part of a generalized sickness response (Dantzer et al., 2008), driven partly by inflammatory cytokines (Vgontzas et al., 1997). Administration of a (typhoid) vaccine that causes an increase in interleukin (IL)-6 levels is paralleled by increased fatigue (Harrison et al., 2009). Along the same lines, observational studies show that low-grade inflammation is related to poor subjective health (Christian et al., 2011; Lekander et al., 2004), in which short or poor sleep is an important factor (Ingre et al., 2008; Singh-Manoux et al., 2006; Steptoe et al., 2006). Sleep-related sickness behaviour is believed to have evolved to conserve energy in order to increase recuperation (Dantzer et al., 2008). In addition, possible background contributors, such as age, gender, depression and anxiety, were included.

Methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure Statement
  9. References

A total of 50 subjects participated. The age range was 18–61 years (mean: 43.5 ± 16.4), 59% were females, 68% were married or cohabiting, 30% had a university education and 70% were employed. Other details are given in Table 1. All were recruited through advertisements and contacts. All subjects were apparently healthy and were screened by a physician. The participants received an economic compensation of approximately $180. The Karolinska Institute ethical committee approved the study. All participants gave written informed consent and the study was carried out according to the principles of the Declaration of Helsinki.

Table 1. Mean ± standard deviation for the main variables
 Overall mean ± SD
  1. aTST: actigraphy total sleep time; dTST: diary total sleep time; i: index; KSS: Karolinska Sleepiness Scale; SD: standard deviation; SRH: self-rated health. Clock time is given in decimal notation.

Sleepiness (KSS 1–9)4.1 ± 1.3
iSleep quality (1–5)4.1 ± 0.8
Lights out (h)23.9 ± 1.6
Time of awakening (h)7.7 ± 1.9
Stress/worries at bedtime (1–5)4.5 ± 0.8
dTST (min)446 ± 96
aTST (min)424 ± 84
Stress average (1–9)2.1 ± 1.2
SRH (1–7)5.4 ± 1.3
Weekend/day off (0/1)0.3 ± 0.5
Coffee (cups; range 0–9)2,4 ± 1.81
Nap (0/1)0.26 ± 0.53
Depression3.1 ± 2.6
Anxiety5.7 ± 3.3
Burnout2.6 ± 1.3

Data were collected daily over a period of 6 weeks. At the start, a background questionnaire was completed. Each morning, a sleep diary with sleep quality ratings was completed, and during the day sleepiness and stress were rated every second hour (8 h, 10 h, 12 h, etc.) during wakefulness. Every evening during the study the participants completed a diary reporting form on health symptoms (Fig. 1). During the entire measurement period, the subjects wore an actigraph for sleep recording. The correlation between actigraphy and self-reported sleep duration was relatively high (= 0.51, < 0.001), and the mean total sleep time (TST) was similar (Table 1).

image

Figure 1. Design and analysis of study. Black: sleep; white: time awake. Thick arrows indicate prospective relation between sleep measures and sleepiness. Dashed arrows indicate times of measurement.

Download figure to PowerPoint

Measurements

Every participant completed a self-administered questionnaire, which included questions about social and demographic background, work conditions and working hours, wellbeing, personality and health. Among the demographic variables were age, gender and being employed/unemployed (or studying). In addition, measures of depression and anxiety were obtained from the Hospital Anxiety and Depression scale (Lisspers et al., 1997).

Sleepiness was rated through the Karolinska Sleepiness Scale (KSS) (Åkerstedt and Gillberg, 1990). This scales ranges from 1 (very alert) to 9 (very sleepy, fighting sleep, an effort to keep awake). In the present study a mean was calculated for each day based on the interval between 8 and 22 h.

Actigraphy was recorded using a device from Cambridge Neurotec®, Cambridge, UK, worn on the wrist of the non-dominant hand. The collected data were scored according to the supplier's algorithms.

The Karolinska sleep diary (KSD) (Åkerstedt et al., 1997), completed in the morning, included: bedtime (h), time of awakening (h), sleep latency (h), sleep quality (how did you sleep: 5, very well–1, very poorly), feeling refreshed after awakening (5, completely–1, not at all), calm sleep (5, very calm–1, very restless), did you get enough sleep (5, definitely enough–1, definitely too little), ease of waking up (5, very easy–1 very difficult) and ease of falling asleep (5, very easy–1, very difficult). Four items formed a sleep quality index (SQI): sleep quality, calmness of sleep, ease of falling asleep and sleep throughout the allotted time. Using bedtime, time of awakening and sleep latency, a measure of diary total sleep time (TST) was derived for each day.

For reports of behaviours and states that might affect sleep, several questions in a wake diary were completed before bedtime. Among them were alcohol consumption (yes/no, 0/1), workday or not (yes/no 0/1), number of cups of coffee during the day (with caffeine) and self-rated health (SRH) (how would you rate your state of health for the day: 1–7, very poor–excellent) (Lekander et al., 2004).

Two types of stress ratings were used. One involved rating stress every second hour during each day on a scale from 1 (no stress at all) to 9 (maximum stress) (Wang et al., 2005). The other item was ‘stress/worries at bedtime’ on a scale from 5 (none) to 1 (very much). The latter has been used for selecting stressful and non-stressful nights for comparison of polysomnograph (PSG) recordings between high and low stress periods (Akerstedt et al., 2007). The item was thought to reflect rumination around stress and problems at bedtime, which has been shown to be an important factor in insomnia (Harvey et al., 2005).

Statistical analysis

The design of the present study involved an analysis of the longitudinal covariation between variables; for example, a night of short sleep reported in the morning predicting reported sleepiness during the rest of the day (Fig. 1). A multi-level analysis was used to deal with the serial dependencies (42 points of measurement) and the two levels of analysis (between and within groups) (Raudenbush and Bryk, 2002).

All statistical analyses were performed with the statistical package stata version 11 (StataCorp, College Station, TX, USA) (Statacorp, 2003). Linear mixed-effect models (Rabe-Hesketh and Skrondal, 2005) were estimated with the STATA procedure ‘xtmixed’. Fewer than 5% of the data were missing.

To evaluate the effect of individual differences in slope and intercept, a set of linear mixed effect regression models, including a random effect over the intercept to allow for variation between subjects, was estimated for each predictor against sleepiness. This was performed for each predictor separately in a univariate model, as well as simultaneously in a multivariate model. The latter reflects the effect of each variable, with all the other variables held constant. The output for fixed effects includes a mean regression coefficient (β) from the individual coefficients, a mean intercept (‘constant’) with the Y-axis, the inferential Z-statistic and its P-value. In addition, the application of random effects permits the regression coefficients to vary between individuals. Also, the intercept with the Y-axis is permitted to vary. The subject-specific predictions (thin lines in Figs 2 and 3) were calculated adding to the fixed-effects (population average) the contribution of the best linear unbiased predictions (BLUPs) of the random effects, often referred to as a ‘semi-Bayes’ prediction. All analyses were adjusted for the effect of linear time (days). None of these effects became significant, however, and are not entered into the tables.

image

Figure 2. Estimates of regressions between sleepiness and diary TST (dTST) and self-rated health (SRH), respectively. Group mean: thick black line; individual regression lines: thin grey lines; upper graph: sleep duration and sleepiness; lower graph: SRH and sleepiness.

Download figure to PowerPoint

image

Figure 3. Scatterplot of sleepiness versus total sleep time (TST) for every 10th participant.

Download figure to PowerPoint

The main focus of the analysis was the prediction of sleepiness from actigraphy or sleep diary ratings of sleep that ended in the morning, 3-hourly stress ratings during the rest of the day, intake of coffee and napping during the same day, evening rating of health for the same day, and whether or not the day was a working day (Fig. 1). In addition, values of the sleep prior to the one in question was included; in particular, previous sleep quality and TST, as a long or short previous sleep may have affected the sleep in question.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure Statement
  9. References

Table 1 presents the mean and standard deviation (SD) for the included variables. The variability was considerable. For the main variables the ±1 SD ranged between 5 h 10 min and 9 h 2 min for TST, between 4.1 and 6.7 on the Karolinska Sleepiness Scale (KSS) and 2.8 and 5.4 on the SRH. The mean of the longest sleep durations was 612 ± 83 min and that of the shortest was 272 ± 68 min [(mean ± standard error(SE)].

The results from the mixed model analysis are presented in Tables 2 and 3 and Figs 2 and 3. The univariate analysis showed that the major predictors of sleepiness were subjective health during the same day and the duration of sleep on the preceding night, with a higher Z-value for diary TST (dTST) than for actigraphy TST (aTST). Sleepiness increased by 0.36 units for every unit drop in reported health (from the intercept of 6.15 for excellent health). Sleepiness fell with 0.15 units for each hour of reported sleep (0.09 for aTST), with an intercept (constant) for sleepiness at 5.3 units (at TST = 0).

Table 2. Results from univariate mixed model regression predicting sleepiness
SleepinessFixed effect Coeff βSEMY-interceptSEM Z P 95% CI for β
  1. All night sleep variables (including stress/worries) refer to the sleep that is immediately followed during the day by the reports of sleepiness, stress average, coffee and napping. prNap refers to napping the day before; ‘I’ indicates index. Fixed-effect coefficient: the regression coefficient for the fixed effect; CI: confidence interval; SEM: standard error of the mean; Z: Z-value for the regression. Y-intercept; SD random: standard deviation of the intercepts (random effect); ‘pr’: prior sleep; TST: total sleep time; SRH: self-rated health.

  2. P: significance level, where *< 0.05, **< 0.01, ***< 0.001, < 0.0001.

Time of awakening (time)−0.0610.0134.640.164.8***−0.085/−0.036
dTST (h)−0.1500.0135.340.17−11.7***−0.180/−0.121
aTST (h)−0.0910.0115.300.20−8.5***−0.115/0−066
iSleep quality (1–5 high)−0.2310.0305.130.17−8.0***−0.288/−0.174
Stress average (1–9 high)0.0790.0254.030.133.2***0.030/0.127
Worries/stress (1–5 high)−0.1490.0274.850.18−5.4***−0.205/−0.095
Weekend /day off (0/1)−0.1480.0414.220.14−3.6***−0.229/−0.068
Coffee (cups)0.0510.0224.020.142.4**0.009/0.093
Nap (0/1)0.3810.0514.100.147.4***0.200/0.482
prNap (0/1)0.0460.0534.180.141.1−0.057/0.149
SRH (1–7)−0.3640.0176.150.1521.2***−0.398/−0.331
Table 3. Results from multivariate mixed model regression predicting sleepiness
SleepinessFixed effect Coeff βSEMZ P95% CI for β
  1. All night sleep variables (including stress/worries) refer to the sleep that is immediately followed during the day by the reports of sleepiness, stress average, coffee and napping. prNap refers to napping the day before. Fixed-effect coefficient: the regression coefficient for the fixed effect; CI: confidence interval; SEM: standard error of the mean; Z: Z-value for the regression. ‘pr’: prior sleep; TST: total sleep time; SRH: self-rated health.

  2. P: significance level, where *< 0.05, **< 0.01, ***< 0.001. < 0.0001.

Time of awakening (h)0.0600.0173.6***0.029/0.100
dTST (h)−0.1600.0128.9***−0.504/−0.120
iSleep quality (1–5 high)−0.0650.0312.1*−0.200/−0.078
Stress average (1–9 high)0.0070.0250.30.001/0.106
Worries/stress (1–5 high)0.0160.0280.6−0.106/0.015
Weekend (0/1)0.0630.0431.4−0.169/0.017
Coffee (cups)0.0250.0201.2−0.017/0.068
Nap (0/1)0.3610.0495.2***0.258/0.465
SRH (1–7)−0.3330.01917.1***−361/−0.321
Y-intercept60.780.23920.4*** 

The two measures of TST were entered into a separate mixed-model analysis against sleepiness, showing that dTST remained a significant predictor whereas aTST did not (Z-values = 6.7*** and 1.8, respectively, and β-coefficients = −0.14 ± 0.02 and −0.03 ± 0.02, respectively). For further information on the relation between the two TST measures, a mixed-model regression was computed predicting dTST from aTST. The result was highly significant (β = 0.46 ± 0.01, = 32.8***, Y-intercept = 109 ± 12.2). Finally, a correlation between the two across times and individuals yielded an r = 0.61, < 0.001).

To illustrate the relation between dTST and sleepiness in raw data, every 10th participant (=6) was plotted in Fig. 3. The results show good correlations for three participants, modest for two and poor for one.

With respect to the other significant predictors, sleepiness increased with quality of previous sleep and the day being a day off (weekend). Sleepiness decreased with stress during the same day, stress/worries before previous sleep, as well as with earlier time of rising, napping and coffee intake during the same day. Note that coffee intake and napping rise with increasing sleepiness the same day.

In the multivariate analysis only the significant variables from the univariate analyses were retained. In addition, aTST was not included because of the obvious collinearity with dTST and the loss of a few subjects due to malfunctioning actigraphs. In the analysis, SRH, dTST, sleep quality, time of awakening and napping remained significant, with essentially the same regression coefficients as in the univariate analysis, but the Z-value was somewhat reduced.

As there is a possibility that curvilinear relationships may exist, quadratic solutions were also sought, but significant results were obtained only for SRH (β = 0.023 (constant 7.1), SE = 0.009, Z = 2.5, < 0.05), and affected the regression only very marginally.

In order to illustrate the effect of TST on sleepiness at different times of day the mean KSS values for the longest and shortest sleep period are plotted in Fig. 4. The minimum TST showed somewhat higher sleepiness values during the ensuing day, particularly towards the evening. Note that n differs between points in time, and is particularly reduced in the early morning and, to some extent, in the late evening.

image

Figure 4. Mean ± standard error (SE) for sleepiness the day after the longest (continuous line) and shortest (dotted line) total sleep time (TST). Note that n differs between points in time: = 15 at 08:00 h, 33 at 10:00 h, 40 at 22:00 h. For the remainder, n varied between 45 and 48.

Download figure to PowerPoint

The background factors were added (each variable separately) to the multivariate analyses. This yielded for age a coefficient of β = −0.187 (per 10 years), with a standard error of 0.076 and a Z-value of −2.40, with < 0.05. Thus, sleepiness fell by 0.2 units for each 10 years of age, indicating that a 50-year-old person would have approximately 0.6 units lower sleepiness than a 20-year-old. The other significant predictors were affected only marginally. No other background factors showed a significant relation with sleepiness.

Apart from the main purpose of the study, there was also an attempt to analyse if dTST had a relation to sleepiness 1 day later. The univariate analysis did not show a significant effect [β = 0.016 ± 0.014, = 1.2, not significant (NS)].

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure Statement
  9. References

The results show that the day-to-day variation in subjective sleepiness was related to preceding TST and end-of-day SRH. Sleep quality, time of awakening and napping were also associated with daytime sleepiness, but to a lesser extent. All significant results remained after adjustment for background variables.

The effect of (reduced) sleep duration on sleepiness has been demonstrated in many laboratory experiments, acutely (Härmä et al., 1998), and with restriction across several days (Axelsson et al., 2008; Belenky et al., 2003; Van Dongen et al., 2003). In this study, however, the effect is present in a day-to-day pattern of spontaneous, real-life variation of TST and the results represent the 42-day average relation of TST with sleepiness. The tight day-to-day coupling observed over time adds credibility to the notion that sleep duration is coupled directly with the daytime level of sleepiness in a dose–response relationship. Assuming that sleepiness is non-conducive to efficient functioning, the results suggest that at least some of the variations between ‘good’ and ‘bad’ days may be due to variations in sleep duration and sleep quality.

As indicated above, KSS increased by 0.15 units for each hour of reduced sleep. This means that a full 8 h of sleep corresponds to 1.2 KSS units. It should be emphasized that the standard deviation of TST was 96 min, around a mean of 446 min. Thus, 64% of the values fell between 350 and 542 min for the average individual; that is, 5.5–9 h. A reduction from 9 to 5.5 h would lead to a increase of sleepiness by approximately 0.55 units. These effects are modest, but the coefficient for TST may have been steeper had the range of variation of TST been greater. That is, if most individuals had occasionally refrained from sleeping (or sleeping for only an hour), the relation between sleep duration and sleepiness would very probably have been considerably steeper. Thus, the present results should mainly apply to average, non-pathological and non-extreme sleep patterns. Sleep quality was a relatively strong predictor in the univariate analysis, but when TST was entered much of this relation disappeared. There was apparently considerable overlap between the two variables.

Actigraphy data support the conclusions based on the self-report data, and the relation between the two measures of sleep duration was considerable. The reason why actigraphy showed a somewhat weaker relation to sleepiness than self-reports may be because of difficulties in recognizing quiet wakefulness from sleep, particularly in real-life settings (Sadeh, 2011), and real-life settings offer more opportunities for lying-in/snoozing than does the well-controlled laboratory study.

Among other sleep-related variables, the univariate effect of weekend disappeared when TST was entered into the analysis, suggesting that it is sleep duration that is active in the reduced sleepiness during the weekend. Nevertheless, the results support the notion of lower sleepiness during days off—on average, 0.36 KSS units. Time of awakening had a small but significant effect, independent of sleep duration, suggesting that a longer time since awakening has a negative effect on alertness, as demonstrated in field studies of acute early rising (Ingre et al., 2004). Note that earlier awakening leads automatically to longer time awake before any given point of measurement of sleepiness. Early rising also leads to a shortening of sleep duration, as bedtime is normally not adjusted to compensate for the early rising (Ingre et al., 2008). However, to our knowledge, the effect of early rising while controlling for sleep duration has not been investigated previously. Apart from early rising, the effect of time awake on sleepiness has been demonstrated previously in studies of sleep deprivation (Fröberg et al., 1975). Lagged TST (that is, TST from 2 nights before the day of rated sleepiness) was not significant in the univariate analysis, thus giving no support for the possibility to bank sleep (Rupp et al., 2009) in the present setting. Napping, finally, was related somewhat strongly to sleepiness on the same day. While napping may be seen as an effective countermeasure to sleepiness (Horne et al., 2008), in this study it seemed, rather, to appear as a response to sleepiness. A study of the sleepiness reduction effects of napping would need another design than that of the present study.

The second major predictor of daily variation in reported sleepiness was daily variations of subjective health. This is a new finding, to the best of our knowledge, but relates to an extensive literature showing that sleepiness (or fatigue) is increased in a number of clinical health problems. Previous findings show that inflammatory cytokines such as IL-6 or tumour necrosis factor-alpha, both of which are strongly sleep-inducing (Krueger, 2008), are also coupled to subjective health perception (Christian et al., 2011; Lekander et al., 2004). Notably, subjective health is measured most often as the general state of health across long time-periods, but in this study was adapted to reflect day-to-day variations in perceived health.

The cytokine discussion above implies SRH as the cause of increased sleepiness. Even if SRH was measured in the evening in the present study, an infection may have been present during the day, causing an increase in sleepiness. Conversely, there is also a possibility that sleepiness (or reduced alertness) during the day (because of sleep loss or for other reasons) is interpreted as a reduction in perceived health. It is possible that the sleepiness ratings, in themselves, may also have produced some common method bias by linking morning ratings of sleep and sleepiness to evening ratings of SRH. However, one must consider that the natural causes of sleepiness, such as short TST, early awakening and poor sleep quality, were all adjusted for in the present analyses. Thus, that chain of causation seems less likely. However, the present design does not permit a conclusion to be drawn on the direction of causation, and the sleepiness/SRH link must be interpreted with caution.

The two stress indicators were significant univariate predictors, but lost their significant contribution when TST was entered. Small amounts of day-to-day stress cause small reductions of sleep efficiency, increased time to slow wave sleep (SWS) onset and reduced reported sleep quality (Akerstedt et al., 2007). Thus, it seems reasonable to attribute the loss of significant contribution of stress in the multivariate analysis to TST and sleep quality.

Apart from the sleep-related determinants of sleepiness, there are still considerable parts of the variance that will affect sleepiness and may have done so in the present study. Monotony is clearly one such factor, as evidenced in studies of, for example, driving-induced sleepiness during real road driving (Sandberg et al., 2011), or simply being active as opposed to inactive. The latter issue does not seem to have been studied systematically.

The present study has several limitations. One concerns the convenience type of the sample. However, it was a study of intra-individual covariation between variables, which should reduce the confounding effects from the selection procedure. The present group was essentially healthy, without any insomniacs or clinically depressed individuals. Thus, the present results are valid only for healthy individuals, from young adults to early retirement age.

Another limitation is that the present study was naturalistic, even if prospective, and it is difficult to attribute cause without experimentation and randomization. Conversely, the interest in the present study was focused specifically on the ecologically valid day-to-day spontaneous covariation. An experimental design would have answered another question and would not have been realistic across a longer period of time, such as the present 42 days. Even if aTST yielded similar links to sleepiness as did dTST, it is still possible that some common method bias had been present. However, the sleep ratings were made upon awakening while the sleepiness ratings were made later during the day (apart from one rating upon awakening). This should have made the influence of a mind-set of the day less likely.

In summary, the results of the present study suggest that, on average, TST will predict sleepiness during the rest of the day across several weeks. Subjective health during the day appears to be an even more important factor, but needs studies with a different design to identify cause and effect.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure Statement
  9. References

This study was supported by the Swedish Science Council, the Swedish Council for Social Sciences and Working Life and Stockholm Stress Center.

Disclosure Statement

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure Statement
  9. References

TÅ has received support from Philips, AstraZeneca and Johnson&Johnson. No other authors have any conflicts of interest to declare.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure Statement
  9. References
  • Åkerstedt, T. and Gillberg, M. Subjective and objective sleepiness in the active individual. Int. J. Neurosci., 1990, 52: 2937.
  • Akerstedt, T., Kecklund, G. and Axelsson, J. Impaired sleep after bedtime stress and worries. Biol. Psychol., 2007, 76: 170173.
  • Åkerstedt, T., Hume, K., Minors, D. and Waterhouse, J. Good sleep—its timing and physiological sleep characteristics. J. Sleep Res., 1997, 6: 221229.
  • Axelsson, J., Kecklund, G., Akerstedt, T., Donofrio, P., Lekander, M. and Ingre, M. Sleepiness and performance in response to repeated sleep restriction and subsequent recovery during semi-laboratory conditions. Chronobiol. Int., 2008, 25: 297308.
  • Belenky, G., Wesensten, N. J., Thorne, D. R. et al. Patterns of performance degradation and restoration during sleep restriction and subsequent recovery: a sleep dose–response study. J. Sleep Res., 2003, 12: 112.
  • Christian, L. M., Glaser, R., Porter, K., Malarkey, W. B., Beversdorf, D. and Kiecolt-Glaser, J. K. Poorer self-rated health is associated with elevated inflammatory markers among older adults. Psychoneuroendocrinology, 2011, 36: 14951504.
  • Dantzer, R., O'Connor, J. C., Freund, G. G., Johnson, R. W. and Kelley, K. W. From inflammation to sickness and depression: when the immune system subjugates the brain. Nat. Rev., 2008, 9: 4656.
  • Dement, W. C. and Carskadon, M. A. Current perspectives on daytime sleepiness: the issues. Sleep, 1982, 5: 5666.
  • Dijk, D. J., Duffy, J. F. and Czeisler, C. A. Circadian and sleep–wake dependent aspects of subjective alertness and cognitive performance. J. Sleep Res., 1992, 1: 112117.
  • Dijk, D. J., Groeger, J. A., Stanley, N. and Deacon, S. Age-related reduction in daytime sleep propensity and nocturnal slow wave sleep. Sleep, 2010, 33: 211223.
  • Fröberg, J. E., Karlsson, C. G., Levi, L. and Lidberg, L. Psychobiological circadian rhythms during a 72-hour vigil. Försvarsmedicin, 1975, II: 192201.
  • Goel, N., Rao, H., Durmer, J. S. and Dinges, D. F. Neurocognitive consequences of sleep deprivation. Semin. Neurol., 2009, 29: 320339.
  • Härmä, M., Suvanto, S., Popkin, S., Pulli, K., Mulder, M. and Hirvonen, K. A dose–response study of total sleep time and the ability to maintain wakefulness. J. Sleep Res., 1998, 7: 167174.
  • Harrison, N. A., Brydon, L., Walker, C. et al. Neural origins of human sickness in interoceptive responses to inflammation. Biol. Psychiatry, 2009, 66: 415422.
  • Harvey, A. G., Tang, N. K. and Browning, L. Cognitive approaches to insomnia. Clin. Psychol. Rev., 2005, 25: 593611.
  • Horne, J., Anderson, C. and Platten, C. Sleep extension versus nap or coffee, within the context of ‘sleep debt’. J. Sleep Res., 2008, 17: 432436.
  • Ingre, M., Kecklund, G., Åkerstedt, T. and Kecklund, L. Variation in sleepiness during early morning shifts: a mixed model approach to an experimental field study of train drivers. Chronobiol. Int., 2004, 21: 973990.
  • Ingre, M., Åkerstedt, T., Peters, B., Anund, A. and Kecklund, G. Subjective sleepiness, simulated driving performance and blink duration: examining individual differences. J. Sleep Res., 2006, 15: 4753.
  • Ingre, M., Kecklund, G., Akerstedt, T., Soderstrom, M. and Kecklund, L. Sleep length as a function of morning shift-start time in irregular shift schedules for train drivers: self-rated health and individual differences. Chronobiol. Int., 2008, 25: 349358.
  • Jewett, M. E., Dijk, D.-J., Kronauer, R. E. and Dinges, D. F. Dose–response relationship between sleep duration and human psychomotor vigilance and subjective alertness. Sleep, 1999, 22: 171179.
  • Krueger, J. M. The role of cytokines in sleep regulation. Curr. Pharm. Des., 2008, 14: 34083416.
  • Landrigan, C. P., Rothschild, J. M., Cronin, J. W. et al. Effect of reducing interns' work hours on serious medical errors in intensive care units. N. Engl. J. Med., 2004, 351: 18381848.
  • Lekander, M., Elofsson, S., Neve, I.-M., Hansson, L.-O. and Unden, A.-L. Self-rated health is related to levels of circulating cytokines. Psychosom. Med., 2004, 66: 559563.
  • Lisspers, J., Nygren, A. and Söderman, E. Hospital Anxiety and Depression Scaale (HAD): some psychometric data for a Swedish sample. Acta Psychiatr. Scand., 1997, 96: 281286.
  • McCrae, C. S., McNamara, J. P., Rowe, M. A. et al. Sleep and affect in older adults: using multilevel modeling to examine daily associations. J. Sleep Res., 2008, 17: 4253.
  • Mezick, E. J., Matthews, K. A., Hall, M. et al. Intra-individual variability in sleep duration and fragmentation: associations with stress. Psychoneuroendocrinology, 2009, 34: 13461354.
  • Philip, P. and Akerstedt, T. Transport and industrial safety, how are they affected by sleepiness and sleep restriction? Sleep Med. Rev., 2006, 10: 347356.
  • Rabe-Hesketh, S. and Skrondal, A. Multilevel and Longitudinal Modeling Using Stata, 1st edn. Stata Press, Texas, 2005.
  • Raudenbush, S. W. and Bryk, A. S.. Hierarchical Linear Models. Applications and data analysis methods, 2nd edn. Sage Publications Inc., Thousand Oaks, CA, USA, 2002.
  • Rupp, T. L., Wesensten, N. J., Bliese, P. D. and Balkin, T. J. Banking sleep: realization of benefits during subsequent sleep restriction and recovery. Sleep, 2009, 32: 311321.
  • Sadeh, A. The role and validity of actigraphy in sleep medicine: an update. Sleep Med. Rev., 2011, 15: 259267.
  • Sagaspe, P., Taillard, J., Akerstedt, T. et al. Extended driving impairs nocturnal driving performances. PLoS ONE, 2008, 3: e3493.
  • Sandberg, D., Anund, A., Fors, C. et al. The characteristics of sleepiness during real driving at night—a study of driving performance, physiology and subjective experience. Sleep, 2011, 34: 13171325.
  • Singh-Manoux, A., Martikainen, P., Ferrie, J., Zins, M., Marmot, M. and Goldberg, M. What does self rated health measure? Results from the British Whitehall II and French Gazel cohort studies. J. Epidemiol. Commun. Health, 2006, 60: 364372.
  • Söderström, M., Ekstedt, M. and Åkerstedt, T. Weekday and weekend patterns of diurnal cortisol, activation and fatigue among people scoring high for burnout. Scand. J. Work Environ. Health, 2006, 2: 3540.
  • Statacorp. Stata Statistical Software, release 8.0. StataCorp LP, College Station, TX, USA, 2003.
  • Steptoe, A., Peacey, V. and Wardle, J. Sleep duration and health in young adults. Arch. Intern. Med., 2006, 166: 16891692.
  • Teixeira, L., Lowden, A., Moreno, C. R. et al. Work and excessive sleepiness among Brazilian evening high school students: effects on days off. Int. J. Occup. Environ. Health, 2010, 16: 172177.
  • Van Dongen, H. P., Maislin, G., Mullington, J. M. and Dinges, D. F. The cumulative cost of additional wakefulness: dose–response effects on neurobehavioral functions and sleep physiology from chronic sleep restriction and total sleep deprivation. Sleep, 2003, 26: 117126.
  • Vgontzas, A. N., Papanicolaou, D. A., Bixler, E. O., Kales, A., Tyson, K. and Chrousos, G. P. Elevation of plasma cytokines in disorders of excessive daytime sleepiness: role of sleep disturbance and obesity. J. Clin. Endocrin. Metab., 1997, 82: 13131316.
  • Wang, J., Rao, H., Wetmore, G. S. et al. Perfusion functional MRI reveals cerebral blood flow pattern under psychological stress. Proc. Natl Acad. Sci., 2005, 102: 1780417809.