The impact of returning to a daytime schedule on sleep, performance and mood after simulated fixed and rotating split shift schedules

Due to growing demands for round-the-clock services, working hours outside of the standard working week are becoming increasingly common, with approximately 1.5 million Australians engaging in shift work (1). Most types of shift work involving night work or long work hours lead to excessive sleepiness and reduced alertness (2, 3). These factors contribute to reduced physical and psychological well-being, lower productivity, and increased risk for accidents (2-5).

Adverse outcomes are mainly a result of circadian misalignment and sleep loss associated with shift work and extended time-on-shift (6). When shift workers are required to work long, overnight shifts, a discord between the sleep-wake schedule and internal circadian rhythm occurs (7). Sleep quantity and quality is substantially reduced during shift workers’ off-duty period, especially when this occurs during the day, due to an increased circadian drive for wake (8). Sleep restriction can occur as a result, causing cumulative decreases in alertness and performance during periods of wake, particularly at night (9). The culmination of homeostatic sleep pressure and the circadian nadir in the early hours of the morning can lead to increased sleepiness and reduced alertness whilst on-shift (10). This is a concern for safety-critical work that occurs over 24h, such as emergency services and healthcare industries. Furthermore, the body is exposed to chronic stress whilst attempting to adjust to new working hours and changing back to regular sleep-wake hours (11), such that sleep and performance may be impaired in the days following a period of shift work.

A number of industries utilise split shift schedules, including transport, maritime and healthcare industries. A benefit of split shift schedules is more frequent sleep opportunities, which should result in a reduction to number of hours awake, and consequently, reduced sleep pressure. Split shift schedules also allow for more nocturnal sleep opportunities compared to long, overnight shifts such as those in 12h on / 12h off shift schedules. Split shift schedules may have start times that are fixed (i.e. the same start times each day such as in 6h on / 6h off shift schedules), or rotating (i.e. start times that rotate over days such as in 8h on / 8h off shift schedules). Several field studies have investigated performance and sleep during split shift schedules (12-18). The majority of studies have found that shifts with 4h on / 8h off schedules appear to be superior in terms of reducing sleepiness, fatigue and performance decrements compared to 6h on / 6h off and 8h on / 8h off shift schedules (19). However, this shift schedule requires increased staff numbers and results in more shift handovers, and therefore may be unviable in some industries. Overall, studies regarding sleep and performance in split shift schedules utilising two crews of workers (such as 6h on / 6h off and 8h on / 8h off) focus on the maritime industry, consist mostly of male samples and lack objective sleep and performance measures. There is also a paucity of information regarding how sleep and performance recover when switching back to a day schedule following a period of split shifts.

This is important to investigate, as there is some evidence to suggest that a recovery period of sleep following a period of shift work can be detrimental rather than beneficial. Van Dongen, Belenky, and Vila (20) assessed psychomotor vigilance performance following a 34h recovery period, which included a total of 20h time-in-bed (TIB), between two cycles of simulated day or night duty. Performance following recovery stabilised in the day duty condition, but continued to decline in the night duty condition. This could be due to those in the night duty condition accumulating deficits that were not recoverable during the limited recovery period, and/or due to experiencing difficulties adapting back to a day schedule, resulting in a reduced ability to utilise the recovery period.

A recent study by Jackson, Banks, and Belenky (21) found differing results in terms of recovery when comparing long shifts (either in the night or the day time) with split shift schedules. Firstly, night-time sleep and split sleep conditions had significantly more total sleep time (TST) than those in the daytime sleep condition during the experimental period. When switching back to a night-time sleep schedule, the split sleep condition had significantly increased TST compared to the daytime sleep condition. This suggests that participants in the daytime sleep condition had more difficulty switching back to a night-sleep schedule. Split shift schedules may therefore be preferable to daytime sleep schedules in terms of ability to utilise a nocturnal sleep opportunity as recovery. Further, it may be that in split shift schedules, where a portion of the sleep is during the biological night, minimal circadian adaptation (or phase change) is likely to occur while on the schedule, resulting in better sleep when switching back to a night-time sleep schedule. Similar arguments have been made for fast-rotating shifts, which, in avoiding circadian adaptation, may result in better recovery on days off (22).

However, it should be noted that shift workers may show no circadian adaptation even on a fixed 12h on / 12h off schedule with consolidated daytime sleep opportunities for several days (23). Importantly, simply altering sleep time by as little as one hour can have flow-on effects for performance. On the day following the switch to Daylight Saving Time in which one hour is lost, workers have been found to sleep 40 minutes less than on other days, and sustain more workplace injuries (24). Therefore, understanding whether, in the absence of any expected circadian adaptation, changes in sleep timing may result in impaired recovery is important when evaluating split shift schedules. Unfortunately, the Jackson et al. study (21) described above did not report performance and mood data during recovery days, and therefore the effect of switching back to a day schedule remains unknown.

Since research suggests that split sleep-wake opportunities provide comparable benefits for performance to a consolidated sleep if TST per 24h is maintained (25, 26), split sleep may not cause sleep loss during a split shift roster. That is, in split shift schedules that are designed to maintain 24h TST, deficits may not accumulate during the shift period. Given that changes in the timing of sleep per se can result in impairment, it is still not clear whether, even under these conditions, impairment may be seen during return to daytime schedules following the roster period. The implication for shift workers is that if recovery is not achieved before a subsequent period of shift work begins, an accumulation of deficits can exacerbate safety risks. Therefore, the current study aimed to investigate whether (a) fixed and rotating split shift schedules with 20h time in bed (TIB) per 48h minimised cumulative deficits in sleep, performance and mood; and (b) returning to a daytime schedule following these shift schedules had a negative impact on sleep, performance and mood.

Overall this study found that there were no differences in waking function between conditions during the shift schedule period, with the exception of positive affect, which was lower in the Rotating condition. Whilst PVT performance remained stable during the shift schedule period compared to baseline, deficits in positive affect, subjective sleepiness and negative affect accumulated over the shift schedule period. The key differences observed in sleep during the shift schedule period were that the Fixed B condition had significantly better sleep than the Fixed A and Rotating conditions, and all conditions had significantly less Stage R sleep compared to BL. Upon RTDS, subjective sleepiness and positive affect returned to BL levels during RTDS, but negative affect remained above BL levels. When switching back to a consolidated, nocturnal sleep, the Fixed B condition had less N3 sleep than at BL. For all conditions, SOL, N2 and N3 latencies were longer than at BL upon RTDS.

Deficits in PVT performance did not accumulate over the shift schedule period, and PVT performance was not affected during the RTDS. While analyses of cumulative performance reported in this paper do not include the potential performance impairment at individual time points during the study, these findings are similar to previous studies, which have found that if adequate TST per 24h is maintained during a split shift schedule, cognitive function is comparable to that after consolidated periods of sleep (21, 25, 26). Subjective sleepiness on the other hand was increased during schedule days for all conditions. However, daily mean scores did not exceed five out of nine, which is similar to scores observed in a previous study on split shifts (21). This score reflects the descriptor “neither alert nor sleepy”. Jackson et al. (21) did not report subjective sleepiness after the split shift period, however the current study found that subjective sleepiness returned to BL levels during RTDS.

Although negative affect increased during the shift schedule period and remained higher than BL levels when switching back to a day schedule, overall scores were low throughout the study and did not exceed an average score of 13. This score falls within the range of normative PANAS scores (16 ± 5.9) observed in a large non-clinical sample (37). Positive affect was significantly lower than BL during the shift schedule period for all conditions. Furthermore, positive affect was lower in the Rotating group compared to the Fixed A and Fixed B schedules, but all conditions were within the normal range, and returned to BL levels when back on a day schedule. A consideration when interpreting changes in mood is that social factors differ in a laboratory, which may have contributed to the observed fluctuations in mood (38, 39). For example, social support outside of the laboratory may keep mood more positive whilst on shift (40). Conversely, social stressors in a real-world setting may lead to further decrements to mood (40) as opposed to inside the laboratory.

The Fixed B condition appeared to be better in terms of sleep quantity, with increased sleep time, sleep efficiency and reduced wake after sleep onset during the shift schedule period compared to the Fixed A and Rotating conditions. These findings differ to a previous study, which found that the two crews on a simulated 6h on / 6h off split shift schedule did not differ in sleep quantity (12). However these disparities may be due to methodological variations, as shift times differed and included a rotation to the alternative schedule during the middle of the simulation. In addition, sleep data were obtained via subjective sleep diaries in the Eriksen study (12). In the current study, the Fixed B condition subsequently had less N3 sleep upon RTDS compared to during BL and the shift schedule period. This is consistent with the reduced sleep disruption throughout the shift schedule, resulting in less accumulation of sleep pressure when switching back to a day schedule (41, 42). In addition, the Fixed B condition had a sleep opportunity that ended 8h prior to the first nocturnal sleep back on a day schedule; therefore less prior wake may have contributed to the reduction in N3 sleep during RTDS. Overall, it appears that participants in the Fixed B schedule slept better during the shift schedule period compared to the other conditions. It is important to note when scheduling workers on a 6h on / 6h off shift schedule that the two rotations of the schedule (i.e. Fixed A versus Fixed B) may not be equal in terms of sleep loss. On the whole, these results support the assumption that split shift schedules with adequate TIB per 24 hours minimise the accumulation of sleep and performance deficits over shift schedule days. However, simply switching back to a consolidated nighttime sleep led to changes in the timing of sleep.

All conditions had significantly longer sleep onset latencies, N2 latency, and N3 latency during RTDS. This may be due to a lack of accumulated sleep pressure making it more difficult to sleep when switching back to a nocturnal sleep schedule (41, 42). Although this means that the schedules succeeded in reducing the accumulation of sleep pressure, a longer sleep onset latency during recovery may have implications when sleep opportunities are less than the 10h allowed in the study. Participants obtained approximately 8h sleep per 10h sleep opportunity on RTDS nights. This long sleep opportunity may not be achievable in a real-world setting which may require workers to commute in between periods of split shift work (40), thus resulting in curtailed opportunities for recovery. Delayed sleep onset could also potentially cause sleep-related anxiety or poor mood (43, 44). Therefore it is important for shift workers to be aware of the fact that experiencing difficulties falling asleep in subsequent days may be due to changing the timing of sleep rather than being indicative of a sleep disorder such as insomnia.

The success of obtaining adequate sleep during split shifts in a real-world setting would also be heavily dependent upon factors including an optimal sleep environment and adhering to shift start and finish times (19). Sleep environments in the laboratory are optimal (e.g. lack of noise, light and social interferences) compared to in the real world (45). Sleep duration may be reduced in operations where sleeping occurs in ship cabins, trains or in emergency service situations such as bushfire fighting, compared to in sleeping areas at home or on layover (19). In the event that sleep quantity and quality during split shift periods is hindered, a situation of sleep restriction would occur, especially during daytime sleep opportunities (11, 46).

Indeed, a recent meta-analysis showed that participants on 8h on / 8h off shift schedules had an average of 5.6h sleep per 24h, while individuals on 6h on / 6h off rosters averaged 5.9h sleep per 24h (19). This is in contrast to the current laboratory study, where participants in the 8h on / 8h off Rotating schedule had an average TST of 7.2h per 24h during the shift schedule period, and average TST of 7.6h and 8.1h on the 6h on / 6h off schedules (Fixed A and Fixed B respectively). Previous studies have found that performance deficits begin to accrue with 7h TIB per 24h for seven nights, and fail to recover to BL levels during subsequent days with 8h TIB opportunities per night (47). Therefore, further studies are needed to ascertain the effects of switching back to a day schedule following split shift schedules when sleep is restricted, and the ability to utilise ‘recovery’ sleep periods in these situations.

It is also important to consider that time of day fluctuations were not captured within the current analyses, which aimed to observe the cumulative cost of rotating and split shift schedules rather than specific time of day effects. However, this does not mean the shift schedules are devoid of periods of performance decrements during circadian low points (7). Especially outside of the laboratory when sleep may be restricted, the build-up of cumulative performance decrements across days may be intensified during the early morning hours (48). Another consideration is that split shift schedules result in an increased number of wake-ups, which are associated with sleep inertia (49). Furthermore, the current participants were sleep-satiated, healthy, young adults. Therefore results may differ in actual shift workers, who are likely to have an increased risk of health issues including metabolic disorder, and potentially engage in unhealthy behaviours such as smoking which can affect sleep (50). Given that some effects were very close to reaching statistical significance, it is possible that increasing the sample size may lead to main effects being detected for condition. For example, the main effect of condition for TST was p=0.062. It is possible, based on patterns emerging in the means, that Fixed B may have obtained significantly less TST if the sample size were larger. Further studies with a larger sample size are needed to confirm this pattern.

The present study shows that the trialled fixed and rotating shift schedules did not produce accumulation of sleep and performance deficits. Switching back to a day schedule following these schedules led to increased sleep latencies and did not improve negative mood. However, it is known that time needed to recover depends on the amount of cumulative sleep loss during a duty period (47), and that sleep environments outside of the laboratory may be sub-optimal. Therefore, further studies are needed to determine whether returning to a daytime schedule acts as a recovery opportunity when sleep is restricted throughout and following a period of split shift work.