We studied the characteristics of sleep and alertness in SWD in various shifts and during free time utilizing information on sleep quantity and quality, sleepiness, and alertness. To our knowledge, this is the first study to investigate SWD in naturalistic morning, evening, and night shifts using both actigraphy and sleep diaries, or to verify disturbed sleep and wake patterns typical to the disorder, as required by the ICSD-3. To ensure that SWD symptoms were related to the shift work schedule, we only qualified as SWD cases those individuals whose questionnaire-based symptoms related to shifts and did not occur in relation to holidays. The results of the field data showed that morning and night shifts induced SWD-related symptoms, and that symptoms during morning shifts in particular seemed to differentiate shift workers with and without SWD. In addition, our results point to poorer recovery from shift work in SWD. This was indicated by greater sleepiness during free time and less compensatory and light sleep on days off.
Insufficient sleep
Earlier research on shift work has shown that shift work per se reduces TST related to night and early morning shifts (Sallinen and Kecklund
2010). Short sleep was also common in this study. The objective daily TST was ≤ 6 h among all the participants related to night shifts, and among those with SWD, also related to morning shifts.
Although objective measures of sleep length showed no significant differences in relation to any work shift or days off, the subjective daily TST was shorter before morning shifts among participants with SWD than among those without SWD. Previous studies using subjective sleep length measures averaged over different shift types have either associated (Kalmbach et al.
2015) or not associated (Di Milia et al.
2013) a decrease in TST with SWD. Instead of studying TST on a general level, we investigated it in relation to different shift types in real life. Our results add to the literature by indicating that subjectively evaluated TST seems to decrease before morning shifts in association with SWD.
Longer sleep latency, lower objective sleep efficiency, and poorer subjective sleep quality suggest that the quality of sleep remains lower among shift workers with SWD than among those without SWD. This is consistent with studies on shift systems that are distinct from the current study, showing poorer subjective sleep quality among swing-shift workers with SWD during a non-work period (Waage et al.
2009), and poorer subjective sleep efficiency among permanent night workers with SWD during night work periods (Gumenyuk et al.
2014) than among controls. In the current study, those with SWD showed rather good sleep quality, with less than 30 min of sleep latency (Sateia et al.
2017), over 85% sleep efficiency (Astill et al.
2013), and a rather good evaluation of quality of sleep. Despite this, both these objective and subjective measures indicated that shift workers with SWD had a poorer capacity to sleep than those without SWD, with respect to all studied shift types and days off.
Subjective sleep need was significantly greater among shift workers with SWD than among those without SWD. It is thus possible that longer sleep need predisposes to SWD. However, in cases of SWD, increased sleepiness and decreased sleep quality may also result in evaluations of longer individual sleep need. In addition, the sleep debt before morning shifts was greater among those with SWD, while compensatory sleep on days off was longer among those without SWD. This greater sleep debt may have resulted from poorer sleep quality, which can lead to slower recovery from work among individuals with SWD. Similarly, shorter sleep debt before morning shifts and the ability to sleep compensatory sleep on days off could have resulted from better sleep quality and subsequently sufficient recovery during the work period, in addition to days off, among the individuals without SWD.
We observed less light sleep on days off among participants with SWD than among those without SWD, which may reflect the difference between the groups’ compensatory sleep. For example, Jay et al. (
2007) have shown that individuals sleep longer after sleep restriction, and that restriction of nightly recovery sleep reduces light and REM sleep. Although recovery sleep was not restricted in the current study, those with SWD may still have not had sufficient compensatory sleep, and would thus need longer to recover than those without SWD. The amount of SWS does not seem to associate with SWD, and based on our results, the poorer recovery of the participants with SWD is more likely to be related to less efficient sleep.
Further, in the current study, shift workers with SWD scored higher points on the bedtime stress scale than those without SWD in association with all shift types and days off. Interestingly, higher job strain in shift work has shown to associate with difficulties in initiating sleep (Karhula et al.
2013), a symptom characteristic to SWD. Sleep-reactivity (i.e. sensitivity of sleep to stress) has been shown to predict SWD (Kalmbach et al.
2015). If the sleep-reactivity in the current study was higher among those with SWD than among those without SWD, the slightly but significantly higher points in the bedtime stress scale could have contributed to the manifestation of the symptoms of SWD, although those with SWD were rather calm and relaxed.
Alertness and sleepiness
We observed more performance lapses in PVT among participants with SWD than among those without SWD at the beginning of night shifts, whereas subjective sleepiness was greater at the end of night shifts, reaching a high level (KSS > 6) (Åkerstedt et al.
2014) among the participants with SWD. The greatest daily sleepiness was high in relation to night and morning shifts. In fact, subjective sleepiness, either the last 5 min KSS or the greatest daily KSS, was greater among the participants with SWD in relation to all the studied shifts, and showed a similar tendency on days off. Regarding night shifts, our findings are consistent with a laboratory study that showed increased night-time sleepiness in association with SWD among permanent night workers (Gumenyuk et al.
2014). However, findings from laboratory studies on permanent night work cannot be extrapolated to real-life shift work as such. To conclude, the current study supports increased sleepiness in association with SWD among shift workers in several shift types.
Differences in flexibility, chronotype, physical exercise, and caffeine consumption
In this study, shift workers with SWD had poorer flexibility in their sleeping habits, which is supported by one epidemiological study (Flo et al.
2012), but unsupported by another (Di Milia et al.
2013). Furthermore, the prevalence of evening chronotype was higher among participants with SWD. Earlier research has both associated (Asaoka et al.
2013) and not associated (Taniyama et al.
2015) evening chronotype with SWD.
As a stimulant, caffeine can disrupt sleep (Wright et al.
2013). Participants with SWD consumed less caffeinated drinks and engaged in physical exercise more often than those without SWD. Individuals reacting with poor sleep to situational stressors such as shift work may also be sensitive to the sleep-disturbing effects of caffeine (Bonnet and Arand
2003) and may avoid consuming it before bedtime. This may have diminished the consumption of caffeine in the current SWD group. Further, exercise can adjust and improve circadian rhythm and sleep (Schroeder and Colwell
2013). However, if poorly scheduled, exercise can disturb sleep, which may have contributed to the impaired sleep and alertness among those with SWD.
Practical relevance—assessment of SWD
Shift work misaligns circadian and diurnal rhythms, almost inevitably causing sleep disturbances. This is a challenge for most shift workers, not just for those with SWD. Due to inefficiency or rigidity in sleeping, shown by less compensatory sleep and less light sleep on days off, generally poorer sleep quality (according to sleep diaries and actigraphy), and less flexibility in sleeping habits, individuals with SWD may not be able to optimally utilize the recovery periods allowed by their rosters. This can degrade alertness and increase the risk of SWD. In fact, the quality of sleep among those without SWD appeared rather high, and this may be essential for coping with recurring disturbance to circadian rhythms.
Our results support the practice of screening the symptoms in naturalistic settings in association with both work shifts and days off since symptoms can appear at different points of time. In this study, the subjective SWD symptoms were more abundant than the objective findings. However, although the current study offers no suggestion of, for example, threshold values, the subjective quality of sleep and sleepiness appear to be useful measures as a first indication of SWD.
Limitations
We did not diagnose SWD by a clinical evaluation, but used a questionnaire to define the disorder, and field measurements to verify it. We were not able to consider all the established factors affecting insomnia and/or sleepiness in this study. However, health conditions with insomnia symptoms and/or sleepiness also typically induce these symptoms while on holiday. The exclusion of volunteers with significant holiday-related symptoms of insomnia and/or sleepiness was likely to reduce the bias due to the diagnosed and non-diagnosed conditions influencing these symptoms.
We verified a reduction in TST in comparison to that on days off in the field data. And although the ICSD-3 does not require verification of reduced TST after a shift work washout period, examining TST after a longer recovery period could have verified whether the SWD cases’ TST had really decreased. However, subjective shift-related TST decreased by approximately 1–2 h in the SWD group when compared to sleep need, bearing in mind that ‘sleep need’ is not equal to ‘TST after a washout period’.
On days off, the participants were given time between 10:00 and 13:00 to complete the PVT, because vigilance performance has shown to be rather stable during this time (Monk et al.
1997). Since cognitive performance follows circadian rhythm it is possible that it may have changed during these three hours (Mollicone et al.
2010).
Because our aim was to provide a detailed picture of sleep and alertness among shift workers who had a SWD compared to those who had none, we included only clear cases and clear non-cases of SWD based on our definition. This increased the contrast between the groups. In our opinion, the inclusion of those with ‘milder SWD symptoms’ in the SWD group would have blurred the comparison since it is unclear whether or not they actually had SWD.
Those with SWD were younger than those without SWD. Young age has been related to shift work tolerance (Saksvik et al.
2011). This may have diminished some group differences, as older age is known to unfavorably affect sleep and recovery (Costa and Sartori
2007; Kiss et al.
2008). Thus, smaller group differences would likely have become significant in a large sample. Nevertheless, statistical power was limited due to small sample size. Moreover, although age and alternatively chronotype were included in the LMMs, we were not able to adjust other analyses for age or chronotype. Similarly, analyses were not adjusted for other possible confounders. In future studies, larger sample sizes could verify our findings. In addition, caution should be used when generalizing our results to shift systems that were not included in this study, for example permanent night work.
Strengths
We based the definition of SWD on the shift-type specific occurrence of insomnia symptoms and sleepiness. As SWD symptoms should associate with work schedules overlapping the usual time for sleep, and recovery from a schedule can take several days, to qualify as having SWD in our study, the participant had to report decreased SWD symptoms after 2 weeks on holiday. All the participants that our screening method classed as positive for SWD also showed a disturbed sleep and wake pattern via both sleep diaries and actigraphy, as required by the ICSD-3. This is also the first study to comply with the ICSD-3’s requirement of reduced TST, verified after defining SWD. Many of the screening methods previously used to define SWD have not been as suitable as our questionnaire. For example, not all studies on SWD have associated SWD symptoms with work schedules that temporally overlap habitual sleep (Di Milia et al.
2013). Our study is the first to use EEG-based measures of sleep or PVT in a field setting for studying SWD. In addition, our study groups were also comparable in terms of shift characteristics, which reduced potential bias deriving from these factors.