This study compared the impact of split and consolidated sleep/wake schedules on subjective sleepiness during the biological day and biological night. This was achieved using a between-group design involving two forced desynchrony protocols: consolidated sleep/wake and split sleep/wake. Both protocols included 7×28-h days with 9.33h in bed and 18.67h of wake each day. While the consolidated sleep/wake protocol had 1×9.33-h sleep opportunity and 1×18.67-h wake period each day, the split sleep/wake protocol had 2×4.67-h sleep opportunities and 2×9.33-h wake periods each day. For both protocols, subjective sleepiness was measured using the Karolinska Sleepiness Scale every 2.5h during wake. A total of 29 healthy adult males participated, with 13 in the consolidated sleep/wake group (mean age=22.5 yrs) and 16 in the split sleep/wake group (mean age=22.6 yrs). On average, subjective sleepiness during wake periods of the split condition was significantly higher than that during the first half of wake periods of the consolidated condition, but was similar to the level during the second half. These findings were observed for wake periods that occurred during both the biological day and biological night. Previous data have shown that cognitive impairment at night is lower for split schedules than consolidated schedules, but the current data indicate that feelings of sleepiness are greater for split schedules than consolidated schedules for at least half of the time awake. Thus, it should be explained to people operating split sleep/wake schedules that although they may perform well, they are likely to feel sleepy.

The accuracy of sleep/wake estimates derived with actigraphy is often dependent on researchers being able to discern non-wear times from sleep or quiescent wakefulness when confronted by discrepancies in a sleep log. Without knowing when an accelerometer is being worn, non-wear could be inferred from periods of inactivity unlikely to occur while in bed. Data collected in our laboratory suggest that more than 50% of inactive periods during time in bed are <8 min in duration. This duration may be an appropriate minimum threshold for routine non-wear classification during self-reported wake. Higher thresholds could be chosen to derive non-wear definitions for self-reported bedtimes depending on the desired level of certainty. To determine non-wear at thresholds of 75%, 95% and 99%, for example, would require periods of inactivity lasting ≥18 min, ≥53 min and ≥85 min, respectively.

Neurobehavioural impairment on the first night shift is often greater than on subsequent night shifts due to extended wakefulness. The aim of the study was to determine whether a 1-h afternoon nap prior to the first night shift is sufficient to produce neurobehavioural performance at levels comparable to the second night shift. Twelve male volunteers (mean age 22.9 years) participated in a laboratory protocol that simulated two 12-h night shifts. A nap preceded the first shift and a 7-h daytime sleep was scheduled between shifts. Neurobehavioural performance and subjective sleepiness measured across each night did not significantly differ between first and second shifts.

Bio-mathematical models that predict fatigue and/or sleepiness have proved a useful adjunct in the management of what has been typically referred to as fatigue-related risk. Codifying what constitutes appropriate use of these models will be increasingly important over the next decade. Current guidelines for determining a safe working time arrangement based on model outputs generally use a single upper threshold and are, arguably, over-simplistic. These guidelines fail to incorporate explicitly essential aspects of the risk assessment process – namely, the inherent uncertainty and variability in human sleep–wake behavior; the non-linear relationship between fatigue, task performance and safety outcomes; the consequence of a fatigue-related error and its influence on overall risk; and the impact of risk mitigation or controls in reducing the likelihood or consequence of a fatigue-related error. As industry and regulatory bodies increasingly move toward performance-based approaches to safety management, any fatigue risk management system that includes a bio-mathematical model should specify what exactly is measured by the model, and how the model can be used in the context of a safety management system approach. This will require significant dialog between the various parties with an interest in bio-mathematical models, i.e. developers, vendors, end-users, and regulators.

The impact of sleep restriction on the endogenous circadian rhythms of hunger and satiety were examined in 28 healthy young men. Participants were scheduled to 2 × 24-h days of baseline followed by 8 × 28-h days of forced desynchrony during which sleep was either moderately restricted (equivalent to 6 h in bed/24 h; n = 14) or severely restricted (equivalent to 4 h in bed/24 h; n = 14). Self-reported hunger and satisfaction were assessed every 2.5 h during wake periods using visual analogue scales. Participants were served standardised meals and snacks at regular intervals and were not permitted to eat ad libitum. Core body temperature was continuously recorded with rectal thermistors to determine circadian phase. Both hunger and satiety exhibited a marked endogenous circadian rhythm. Hunger was highest, and satiety was lowest, in the biological evening (i.e., ~17:00–21:00 h) whereas hunger was lowest, and satiety was highest in the biological night (i.e., 01:00–05:00 h). The results are consistent with expectations based on previous reports and may explain in some part the decrease in appetite that is commonly reported by individuals who are required to work at night. Interestingly, the endogenous rhythms of hunger and satiety do not appear to be altered by severe—as compared to moderate—sleep restriction.

Two SOM ANN approaches were used in a study of Australian railway drivers (RDs) to classify RDs' sleep/wake states and their sleep duration time series profiles over 14 days follow-up. The first approach was a feature-based SOM approach that clustered the most frequently occurring patterns of sleep. The second created RD networks of sleep/wake/duty/break feature parameter vectors of between-states transition probabilities via a multivariate extension of the mixture transition distribution (MTD) model, accommodating covariate interactions. SOM/ANN found 4 clusters of RDs whose sleep profiles differed significantly. Generalised Additive Models for Location, Scale and Shape of the 2 sleep outcomes confirmed that break and sleep onset times, break duration and hours to next duty are significant effects which operate differentially across the groups. Generally sleep increases for next duty onset between 10 am and 4 pm, and when hours since break onset exceeds 1 day. These 2 factors were significant factors determining current sleep, which have differential impacts across the clusters. Some drivers groups catch up sleep after the night shift, while others do so before the night shift. Sleep is governed by the RD's anticipatory behaviour of next scheduled duty onset and hours since break onset, and driver experience, age and domestic scenario. This has clear health and safety implications for the rail industry.

ABSTRACT The aim of this study was to examine the sleep strategies used in breaks between consecutive shifts. For two weeks, 253 shiftworkers collected data regarding the timing of all shifts and sleeps. The final dataset included 395 between-shift breaks that had a standard duration (i.e. ~16 h). If a break included an entire night, participants had a single sleep on 80–93% of occasions, but if a break occurred predominantly during the daytime, participants had more than one sleep on 41–50% of occasions. These data indicate that the sleep strategy employed is influenced by the time-of-day that a break occurs.

Fatigue is a significant contributor to motor-vehicle accidents and fatalities. Shift workers are particularly susceptible to fatigue-related risks as they are often sleep-restricted and required to commute around the clock. Simple assays of performance could provide useful indications of risk in fatigue management, but their effectiveness may be influenced by changes in their sensitivity to sleep loss across the day. The aim of this study was to evaluate the sensitivity of several neurobehavioral and subjective tasks to sleep restriction (SR) at different circadian phases and their efficacy as predictors of performance during a simulated driving task. Thirty-two volunteers (M±SD; 22.8±2.9 years) were time-isolated for 13-days and participated in one of two 14-h forced desynchrony protocols with sleep opportunities equivalent to 8h/24h (control) or 4h/24h (SR). At regular intervals during wake periods, participants completed a simulated driving task, several neurobehavioral tasks, including the psychomotor vigilance task (PVT), and subjective ratings, including a self-assessment measure of ability to perform. Scores transformed into standardized units relative to baseline were folded into circadian phase bins based on core body temperature. Sleep dose and circadian phase effect sizes were derived via mixed models analyses. Predictors of driving were identified with regressions. Performance was most sensitive to sleep restriction around the circadian nadir. The effects of sleep restriction around the circadian nadir were larger for simulated driving and neurobehavioral tasks than for subjective ratings. Tasks did not significantly predict driving performance during the control condition or around the acrophase during the SR condition. The PVT and self-assessed ability were the best predictors of simulated driving across circadian phases during SR. These results show that simple performance measures and self-monitoring explain a large proportion of the variance in driving when fatigue-risk is high.

Biomathematical models of fatigue can assist organisations to estimate the fatigue consequences of a roster before operations commence. These estimates do not account for the diversity of sleep behaviours exhibited by employees. The purpose of this study was to develop sleep transfer functions describing the likely distributions of sleep around fatigue level estimates produced by a commercial biomathematical model of fatigue. Participants included 347 (18 females, 329 males) train drivers working commercial railway operations in Australia. They provided detailed information about their sleep behaviours using sleep diaries and wrist activity monitors. On average, drivers slept for 7.7 (±1.7)h in the 24h before work and 15.1 (±2.5)h in the 48h before work. The amount of sleep obtained by drivers before shifts differed only marginally across morning, afternoon and night shifts. Shifts were also classified into one of seven ranked categories using estimated fatigue level scores. Higher fatigue score categories were associated with significant reductions in the amount of sleep obtained before shifts, but there was substantial within-category variation. The study findings demonstrate that biomathematical models of fatigue have utility for designing round-the-clock rosters that provide sufficient sleep opportunities for the average employee. Robust variability in the amount of sleep obtained by drivers indicate that models are relatively poor tools for ensuring that all employees obtain sufficient sleep. These findings demonstrate the importance of developing approaches for managing the sleep behaviour of individual employees.

The aim of this study was to compare the quantity/quality of sleep obtained by people living on split and consolidated sleep-wake schedules. The study had a between-groups design, with 13 participants in a consolidated condition (all males, mean age of 22.5yr) and 16 participants in a split condition (all males, mean age of 22.6yr). Both conditions employed forced desynchrony protocols with the activity:rest ratio set at 2:1, but the consolidated condition had one sleep-wake cycle every 28h (9.33+18.67), while the split condition had one sleep-wake cycle every 14h (4.67+9.33). Sleep was assessed using polysomnography. Participants in the split and consolidated conditions obtained 4.0h of sleep per 14h and 7.6h of sleep per 28h, respectively. Some differences between the groups indicated that sleep quality was lower in the split condition than the consolidated condition: the split sleeps had longer sleep onset latency (9.7 vs. 4.3min), more arousals (7.4 vs. 5.7 per hour in bed), and a greater percentage of stage 1 sleep (4.1% vs. 3.1%), than the consolidated sleeps. Other differences between the groups indicated that sleep quality was higher in the split condition than the consolidated condition: the split sleeps had a lower percentage of wake after sleep onset sleep (11.7% vs. 17.6%), and a greater percentage of slow wave sleep (30.2% vs. 23.8%), than the consolidated sleeps. These results indicate that the split schedule was not particularly harmful, and may have actually been beneficial, to sleep. Split work-rest schedules can be socially disruptive, but their use may be warranted in work settings where shiftworkers are separated from their normal family/social lives (e.g., fly-in fly-out mining) or where the need for family/social time is secondary to the task (e.g., emergency response to natural disasters).

The objective of this study was to assess the validity of a sleep/wake activity monitor, an energy expenditure activity monitor, and a partial-polysomnography system at measuring sleep and wake under identical conditions. Secondary aims were to evaluate the sleep/wake thresholds for each activity monitor and to compare the three devices. To achieve these aims, two nights of sleep were recorded simultaneously with polysomnography (PSG), two activity monitors, and a partial-PSG system in a sleep laboratory. Agreement with PSG was evaluated epoch by epoch and with summary measures including total sleep time (TST) and wake after sleep onset (WASO). All of the devices had high agreement rates for identifying sleep and wake, but the partial-PSG system was the best, with an agreement of 91.6% ± 5.1%. At their best thresholds, the sleep/wake monitor (medium threshold, 87.7% ± 7.6%) and the energy expenditure monitor (very low threshold, 86.8% ± 8.6%) had similarly high rates of agreement. The summary measures were similar to those determined by PSG, but the partial-PSG system provided the most consistent estimates. Although the partial-PSG system was the most accurate device, both activity monitors were also valid for sleep estimation, provided that appropriate thresholds were selected. Each device has advantages, so the primary consideration for researchers will be to determine which best suits a given research design.

Extended wakefulness, sleep loss, and circadian misalignment are factors associated with an increased accident risk in shiftwork. Splitting shifts into multiple shorter periods per day may mitigate these risks by alleviating prior wake. However, the effect of splitting the sleep-wake schedule on the homeostatic and circadian contributions to neurobehavioural performance and subjective assessments of one's ability to perform are not known. Twenty-nine male participants lived in a time isolation laboratory for 13 d, assigned to one of two 28-h forced desynchrony (FD) schedules. Depending on the assigned schedule, participants were provided the same total time in bed (TIB) each FD cycle, either consolidated into a single period (9.33 h TIB) or split into two equal halves (2 x 4.67 h TIB). Neurobehavioural performance was regularly assessed with a psychomotor vigilance task (PVT) and subjectively-assessed ability was measured with a prediction of performance on a visual analogue scale. Polysomnography was used to assess sleep, and core body temperature was recorded to assess circadian phase. On average, participants obtained the same amount of sleep in both schedules, but those in the split schedule obtained more slow wave sleep (SWS) on FD days. Mixed-effects ANOVAs indicated no overall difference between the standard and split schedules in neurobehavioural performance or predictions of performance. Main effects of circadian phase and prior wake were present for both schedules, such that performance and subjective ratings of ability were best around the circadian acrophase, worst around the nadir, and declined with increasing prior wake. There was a schedule by circadian phase interaction for all neurobehavioural performance metrics such that performance was better in the split schedule than the standard schedule around the nadir. There was no such interaction for predictions of performance. Performance during the standard schedule was significantly better than the split schedule at 2 h of prior wake, but declined at a steeper rate such that the schedules converged by 4.5-7 h of prior wake. Overall, the results indicate that when the total opportunity for sleep per day is satisfactory, a split sleep-wake schedule is not detrimental to sleep or performance. Indeed, though not reflected in subjective assessments of performance capacity, splitting the schedule may be of some benefit, given its reduction of neurobehavioural impairment at night and its association with increased SWS. Therefore, for some industries that require operations to be sustained around the clock, implementing a split work-rest schedule may be of assistance.

Previous forced desynchrony studies have highlighted the close relationship between the circadian rhythms of core body temperature (CBT) and sleep propensity. In particular, these studies have shown that a "forbidden zone" for sleep exists on the rising limb of the CBT rhythm. In these previous studies, the length of the experimental day was either ultrashort (90 min), short (20 h), or long (28 h), and the ratio of sleep to wake was normal (i.e., 1:2). The aim of the current study was to examine the relative effects of the circadian and homeostatic processes on sleep propensity using a 28-h forced desynchrony protocol in which the ratio of sleep to wake was substantially lower than normal (i.e., 1:5). Twenty-seven healthy males lived in a time-isolation sleep laboratory for 11 consecutive days. Participants completed either a control (n = 13) or sleep restriction (n = 14) condition. In both conditions, the protocol consisted of 2 × 24-h baseline days followed by 8 × 28-h forced desynchrony days. On forced desynchrony days, the control group had 9.3 h in bed and 18.7 h of wake, and the sleep restriction group had 4.7 h in bed and 23.3 h of wake. For all participants, each 30-s epoch of time in bed was scored as sleep or wake based on standard polysomnography recordings, and was also assigned a circadian phase (360° = 24 h) based on a cosine equation fitted to continuously recorded CBT data. For each circadian phase (i.e., 72 × 5° bins), sleep propensity was calculated as the percentage of epochs spent in bed scored as sleep. For the control group, there was a clear circadian rhythm in sleep propensity, with a peak of 98.5% at 5° (~05:20 h), a trough of 64.9% at 245° (~21:20 h), and an average of 82.3%. In contrast, sleep propensity for the sleep restriction group was relatively high at all circadian phases, with an average of 96.7%. For this group, the highest sleep propensity (99.0%) occurred at 60° (~09:00 h), and the lowest sleep propensity (91.3%) occurred at 265° (~22:40 h). As has been shown previously, these current data indicate that with a normal sleep-to-wake ratio, the effect of the circadian process on sleep propensity is pronounced, such that a forbidden zone for sleep exists at a phase equivalent to evening time for a normally entrained individual. However, these current data also indicate that when the ratio of sleep to wake is substantially lower than normal, this circadian effect is masked. In particular, sleep propensity is very high at all circadian phases, including those that coincide with the forbidden zone for sleep. This finding suggests that if the homeostatic pressure for sleep is sufficiently high, then the circadian drive for wakefulness can be overridden. In future studies, it will be important to determine whether or not this masking effect occurs with less severe sleep restriction, e.g., with a sleep-to-wake ratio of 1:3.

Biomathematical models are used in industry to estimate how much sleep people are likely to get on different work patterns, and how efficient and safe people are likely to be at work. Since there is evidence to suggest that individuals respond differently to sleep loss, there has been a recent focus on trying to account for individual differences. One possible approach could use past behaviour to predict future responses to similar working conditions. This study investigated the predictive value of sleep timing and duration data for a particular individual on a break between shifts relative to data from their colleagues. Sleep diaries and wrist actigraphy were collected from 306 international long-haul pilots for at least 2-weeks. Fifty layovers, equivalent in origin and destination, length and timing, were completed twice by individual pilots. Matched layovers done by other pilots (n = 2311) were also identified. Layover periods were analysed for minute-by-minute correspondence of sleep or wake (yes/no), and total sleep time (TST). Using an individual's own data improved concordance by approximately 5% relative to using a large sample of different pilots, and by 10% relative to using a random sample of 50 different pilots. Using an individual's own TST to predict their TST on an equivalent layover yielded an r value of 0.83, compared to r = 0.78 when data from a colleague was used, and r = 0.73 using different pilots in a random sample of equivalent size. The mean difference in TST using pilots' own data was <20 min, compared to <40 min using data from colleagues. However, the confidence limits on these differences were large (up to 8 h). Results suggest that for international pilots on specific layover patterns, knowing the past behaviour of an individual may only represent a modest improvement over knowing the length and timing of a colleague's sleep, when it comes to predicting their sleep behaviour.