Helen Burgess, PhD, Biological Rhythms Research Laboratory, Rush University Medical Center, 1645 W. Jackson Blvd, Suite 425, Chicago, IL 60612, USA. Tel.: 1-(312)-563-4785; fax: 1-(312)-563-4900; e-mail: firstname.lastname@example.org
Short sleep/dark durations are common in modern society. We investigated if exposure to additional evening ambient light, often associated with later bedtimes and short sleep, reduces circadian phase advances to morning bright light. Twelve healthy subjects participated in two conditions that differed in the distribution of sleep before exposure to morning bright light. Subjects had a consolidated 9-h night time sleep opportunity, or a 3-h daytime nap followed by a 6-h night time sleep opportunity, each before morning bright light. Eight of the 12 subjects obtained similar amounts of sleep in both conditions, and yet still showed significant reductions in phase advances with 6-h nights (1.7 versus 0.7 h, P < 0.05). These results suggest that the exposure to additional evening ambient light often associated with short sleep episodes can significantly reduce phase advances to morning light, and may therefore increase the risk for circadian misalignment.
Many people in modern society regularly experience short sleep episodes. In 2010 the National Sleep Foundation found up to 27% of Americans reported obtaining <6 h of sleep on weeknights, which was mostly due to later bedtimes (National Sleep Foundation, 2010). Indeed, more time spent working and commuting combined with a reluctance to curtail social and leisure time contribute to this increasing societal preference for shorter sleep (Knutson et al., 2010). We previously reported that 2 weeks of short sleep episodes (6 h night−1) led to a 50% reduction in phase advances to morning light, compared with 2 weeks of long sleep episodes (9 h night−1; Burgess and Eastman, 2005). As most humans have an endogenous circadian tendency to phase delay (Burgess and Eastman, 2008; Duffy et al., 2011), morning light is essential for producing corrective phase advances to maintain proper alignment with the external 24-h day. By reducing phase advances to morning light, short sleep episodes may increase the risk for circadian misalignment and the associated negative impact on health (Janszky and Ljung, 2008), mood, sleepiness and cognitive performance (Levandovski et al., 2011; Taylor et al., 2008; Yang and Spielman, 2001).
The potential mechanism(s) underlying our initial observation that short nights significantly reduce phase advances to light remain unclear (Burgess and Eastman, 2005). The sleep deprivation associated with short sleep episodes is one potential mechanism by which short sleep episodes reduce phase advances to light (Challet et al., 2001; Mistlberger et al., 1997). In a follow up study in humans, we found that partial sleep deprivation reduces phase advances to morning light (Burgess, 2010), but the effect was modest and could not account for the previously observed 50% reduction (Burgess and Eastman, 2005). An additional cause of the reduction in phase advances to morning light could be the increased exposure to ambient light in the evening, which is often associated with later bedtimes and short sleep episodes. Indeed, previous work demonstrates that room light in the evening can suppress melatonin (Gooley et al., 2011; Santhi et al., 2011) and phase delay the human circadian clock (Burgess and Eastman, 2004). Together these studies suggest that evening light has the potential to reduce phase advances to morning light. Therefore, in this study we investigated the effect of additional ambient evening light exposure while controlling for sleep deprivation, to further determine the mechanism(s) behind how short sleep episodes reduce the circadian response to morning light.
Materials and Methods
Twelve healthy subjects participated (six men, six women, mean age ± SD = 28.5 ± 7.5 years; body mass index 25.1 ± 2.9 kg m−2). They were non-smokers, medication free, consumed moderate caffeine (<300 mg day−1) and alcohol doses (<2 drinks day−1), and reported no medical, psychiatric or sleep disorders as assessed from screening questionnaires (Beck et al., 1961; Butcher et al., 1989; Buysse et al., 1989). A urine drug screen confirmed all subjects were free of common drugs of abuse. No subject was color blind, as determined by the Ishihara test. No subject had worked night shifts or travelled across more than one time zone in the month preceding the study. The self-reported mean (±SD) sleep schedule in the week before the study was 00:23 ± 1.0 to 08:13 ± 1.1 h. Morningness–eveningness was assessed (Horne and Ostberg, 1976), and there were two moderate morning, eight neither and two moderate evening types. Subjects gave written informed consent prior to their participation. The study was approved by the Rush University Medical Center Institutional Review Board.
The study was 48 days long and was a within-subjects design to assess how evening room light exposure influences circadian phase advances to bright morning light, while attempting to control for sleep deprivation (Fig. 1). The study had two parts, and each part had 6 baseline nights at home, a baseline phase assessment in the laboratory, 7 more baseline nights at home, a 3-day phase-advancing protocol and a final phase assessment in the laboratory. The two parts differed in the timing of a 9-h sleep opportunity during the 3-day phase-advancing protocol in the laboratory. In one condition (‘long nights’), subjects had a consolidated 9-h sleep opportunity per night. In the other condition (‘short nights’), subjects had a 3-h daytime nap followed by a 6-h sleep opportunity per night. Thus, the total sleep opportunity was the same in both conditions, but in the short nights condition subjects received 3 h of dark in the middle of each day and 3 additional hours of evening ambient light per night. There was a 9-day break between the two parts of the study, during which subjects returned to their baseline sleep times. Six subjects completed the long nights condition first and six subjects completed the short nights condition first.
Sleep at home
All subjects slept at home except during the 3 days of the phase-advancing protocol when they slept in the laboratory. Subjects were assigned a fixed 8-h home sleep schedule that was within 1 h of their self-reported habitual sleep times. Subjects were instructed to maintain their usual routines at home. Subjects were also required to go outside to receive a minimum of 10 min of morning light to mimic the morning light that many people receive every day. The light exposure had to occur within a 1-h window, starting 30 min after their scheduled wake time. Compliance to the home sleep requirements was ensured with wrist activity monitors and light sensors worn around the neck (Actiwatch-L, Respironics, Bend, OR, USA), as per our previous study (Burgess, 2010).
Each subject participated in four dim light phase assessments in the laboratory to determine their dim light melatonin onset (DLMO), which is a reliable marker of the circadian clock (Lewy et al., 1999; Fig. 1). Methodological details of the phase assessments have been previously described (Burgess, 2010). Briefly, subjects remained awake and seated in dim light (<5 lux, at the level of the eyes, in the direction of gaze; Minolta TL-1 light meter, Ramsey, NJ, USA), and gave a saliva sample every 30 min using Salivettes (Sarstedt, Newton, NC, USA). Subjects were not permitted to consume any alcohol or caffeine after the third baseline day, and were breathalysed at the start of each laboratory stay. Non-steroidal anti-inflammatory drugs were not permitted during the entire study (Murphy et al., 1996).
Saliva samples from the first 11–14 h of the phase assessments were radioimmunoassayed for melatonin by Pharmasan Laboratories (Osceola, WI, USA). The sensitivity of the assay was 0.7 pg mL−1, and intra- and interassay coefficients of variabilities were 12.1 and 13.2%, respectively. A DLMO was calculated for each phase assessment, and defined as the point in time (with linear interpolation) when the melatonin concentration exceeded the mean of five low consecutive daytime values plus twice the standard deviation of these points (Burgess, 2010; Voultsios et al., 1997). A separate threshold was calculated for each melatonin profile, but the thresholds did not vary between conditions (paired t-test, P = 0.18). The phase shift during each condition was calculated as the baseline DLMO minus the final DLMO.
The timing of each phase-advancing protocol in the laboratory was dependent on the timing of each subject’s baseline DLMO, as determined from the previous baseline phase assessment (Fig. 1). Subjects arrived at the laboratory 10.5 h before their baseline DLMO on Day 15, and remained in the laboratory until the end of the final phase assessment. Subjects were not permitted to exercise in the laboratory. In the long nights condition subjects were put to bed in an individual temperature-controlled dark bedroom 1 h before their baseline DLMO. In the short nights condition there was a 3-h nap centered in the middle of each waking day. This nap was unlikely to have altered circadian phase, as neither a 6-h nap or 5 h of bright light exposure at this time phase shift circadian rhythms (Buxton et al., 2000; Dumont and Carrier, 1997). Subjects were also put to bed 2 h after their baseline DLMO, and so compared with the long nights condition subjects received 3 additional hours of ambient light before bedtime each night. Light readings at subjects’ angle of gaze with a Minolta TL-1 light meter (Ramsey, NJ, USA) during these 3 h were <45 lux, on average 18.2 ± 6.0 (SD) lux. Light readings during these 3 h from the light sensors worn around the neck were <84 lux, on average 15.6 ± 7.0 (SD) lux.
In both conditions subjects were awakened 8 h after their baseline DLMO, and exposed to bright intermittent light spanning 3.5 h [mean intensity 4694 ± 883 (SD) lux], 30 min bright light alternating with 30 min ordinary, dim room light <120 lux (measured periodically at angle of gaze with Minolta TL-1 light meter; Ramsey, NJ, USA), as per our previous study (Burgess, 2010). The bright light was timed to begin 8 h after the DLMO to control for the circadian timing of light as per our previous studies (Burgess, 2010; Burgess and Eastman, 2005). The timing of the sleep/dark episode and bright light exposure was advanced by 1 h day−1 over the next 2 days of the phase-advancing protocol. Following each 3-day phase-advancing protocol, subjects had a final phase assessment.
Each subject’s sleep–wake state during each sleep opportunity in the 3-day phase-advancing protocol was assessed with polysomnography to later determine if subjects obtained similar amounts of sleep in both conditions. Sleep was assessed with Vitaport 3 systems with a 12-bit analog-to-digital convertor (Temec Instruments, Kerkrade, the Netherlands). Electroencephalogram (F4/M1, F3/M2, C4/M1, C3/M2, O2/M1, O1/M2), electrooculogram, electromyogram (submental) and electrocardiogram signals were filtered, sampled and stored according to standard recommendations (Iber et al., 2007). The sleep recordings were later scored by a certified sleep technologist according to standard procedures (Iber et al., 2007), and average total sleep time per day was calculated for each condition.
Measures of sleepiness and performance
During the phase-advancing protocol subjects completed the Stanford Sleepiness Scale (SSS; Hoddes et al., 1973), followed by a 5-min Psychomotor Vigilance Test (PVT) on a portable handheld device (Burgess, 2010). The subjects completed the SSS and PVT twice daily: 15 min before bedtime in the long nights condition; and 3 h and 15 min before bedtime in the short nights condition; and within 15 min after their assigned wake time. The reciprocal mean reaction time and fastest 10% reaction times were extracted from each PVT trial as these variables are most sensitive to partial sleep deprivation (Basner and Dinges, 2011).
The average total sleep time in each condition was analysed with a two-way anova with a within-subjects factor Condition (long versus short nights) and a between-subjects factor Order (long versus short nights first). The phase shifts in the DLMO were analysed with a three-way anova with a within-subjects factor Condition (long versus short nights), within-subjects factor Time (circadian phase before and after phase-advancing protocol) and a between-subjects factor Order (long versus short nights first). The SSS and PVT variables were analysed with a three-way repeated-measures anova with a within-subjects factor Condition (long versus short nights), within-subjects factor Time (bedtime versus wake time) and a between-subjects factor Order (long versus short nights first). For all these analyses the Condition main effects and Condition × Time interactions were of most interest as they would indicate if there was a significant difference between the two conditions. Statistical significance for all analyses was determined with two-tailed tests at P < 0.05. Results are reported as means ± SD.
There were no significant order effects in any of the analyses, and so this factor was removed from all anovas. Overall, the subjects received significantly less total sleep in the short nights than long nights condition (436.81 ± 40.68 versus 468.38 ± 48.74 min day−1, Condition main effect: F1,11 = 6.93, P < 0.05). This difference in sleep was reflected in the SSS ratings (Condition × Time interaction: F1,11 = 9.57, P < 0.05), with more sleepiness reported after waking in the short nights (average 3.83 ± 0.88) versus long nights condition (average 3.44 ± 1.11). PVT data from one subject was lost due to equipment failure. In the remaining 11 subjects, the sleep loss in the short nights was not reflected in either PVT variable (both Condition and Condition × Time interactions, P > 0.05), although reaction time performance was significantly worse after wake than before bed (e.g. fastest 10% reaction time before bed average 215.15 ± 32.57 versus after wake average 233.64 ± 38.87, Time: F1,10 = 9.65, P < 0.05). The DLMOs in a representative individual subject are shown in Figs 2 and 3 shows the phase shifts in the two conditions in the entire sample. The DLMO phase advanced significantly less in the short nights than in the long nights condition (1.93 ± 0.98 versus 0.80 ± 0.92 h, Condition × Time interaction: F1,11 = 18.67, P < 0.05). The average reduction in phase advance ranged from 0.03 h to 3.15 h.
Four subjects lost 40 min or more sleep per night in the short nights versus long nights condition. However, the remaining eight subjects obtained similar amounts of sleep in the short and long nights conditions (449.10 ± 34.61 versus 461.32 ± 51.05 min, Condition main effect: F1,7 = 0.89, P > 0.05). Additionally, this subgroup of eight subjects did not show any differences in sleepiness ratings on the SSS between conditions (Condition main effect and Condition × Time interaction, both P > 0.05). Further analysis of these eight subjects indicated that even without significant sleep deprivation, the DLMO still phase-advanced significantly less in the short nights than in the long nights condition (1.67 ± 1.06 versus 0.71 ± 0.82 h, Condition × Time interaction: F1,7 = 13.84, P < 0.05).
The results of this 48-day within-subjects design study indicate that short nights (6 h night−1) can significantly reduce phase advances to morning light as compared with long nights (9 h night−1), even in the absence of significant sleep deprivation. These results confirm earlier reports that short nights can significantly reduce phase advances to morning bright light (Burgess and Eastman, 2005), and extend the original findings by controlling for sleep deprivation while establishing the significance of additional evening ambient light exposure as a potential mechanism by which short nights reduce phase advances to light.
The most likely cause of the reduction in phase advances during the short nights is the additional 3 h of evening ambient light before bedtime in the short nights condition. The human circadian system is sensitive to evening light, as demonstrated by the ability of evening room light to suppress melatonin (Gooley et al., 2011; Santhi et al., 2011). Indeed, this additional evening light in the short nights condition may have reduced the sensitivity of the circadian system to the bright light the next morning (Chang et al., 2011), thereby reducing subsequent phase advances. The evening ambient light may have also phase-delayed the circadian clock, as repeated exposure to dim ambient light in the home environment has been shown to delay circadian phase (Burgess and Eastman, 2004). This phase delay may have counteracted the phase advances to the morning bright light, also accounting for the smaller phase advances to bright light in the short nights. Finally, in this study, the morning bright light started 8 h after the baseline DLMO on the first morning, in an attempt to administer the bright light at the same circadian phase in both conditions (Burgess, 2010; Burgess and Eastman, 2005), and at a time that bright light produces large phase advances (Khalsa et al., 2003). However, with smaller phase advances in the short nights condition, it is possible that the morning bright light was no longer optimally timed on the second and third morning of bright light exposure, and this effect could have also contributed to the smaller phase advances in the short nights condition.
There are other possible causes of the reduced phase advances in the short nights condition. These include the 3-h naps in the center of the waking day in the short nights condition. However, it appears unlikely the naps per se significantly altered the phase advances to light, because a 6-h nap in the center of the waking day has been shown not to significantly change circadian phase (Buxton et al., 2000), just as 5 h of bright light exposure at this time does not significantly change circadian phase (Dumont and Carrier, 1997). Other non-photic zeitgebers, such as changes in posture, arousal and meal times in the laboratory, may have also contributed to the results, but there is little evidence to suggest they can significantly alter circadian phase in humans, with light considered the strongest zeitgeber in humans (Mistlberger and Skene, 2004). A limitation of the study design is that the first 3 h of the long nights consisted of not only dark (no evening ambient light) but also sleep. For this reason, an additional condition of short sleep prefaced by 3 h in the dark ideally would have been included in the study design. Nonetheless, as we controlled for total sleep amount in both conditions, the most likely explanation for our findings is that increased exposure to evening ambient light reduces phase advances to morning light.
These findings help establish how the human circadian system is influenced by and responds to relatively recent trends in voluntary human behavior. These results suggest that the increasing societal trend towards shorter sleep episodes (Knutson et al., 2010) is likely to significantly impact circadian responsiveness to light and therefore predispose short sleepers to circadian misalignment. For example, in the short nights we observed a small average phase advance of 0.8 h in response to a 3-day phase-advancing protocol, which averages to about 0.3 h advance per day. This advance, at least on average, may not be enough to immediately overcome phase delays that commonly occur in the general population after the transition to daylight saving time (Kantermann et al., 2007) or after returning to work following later weekend sleep (Taylor et al., 2008; Yang et al., 2001). While these phase delays are relatively subtle, they can nonetheless have significant negative effects on health (Janszky and Ljung, 2008), mood (Levandovski et al., 2011), sleepiness and cognitive performance (Taylor et al., 2008; Yang and Spielman, 2001; Yang et al., 2001). Thus, short sleepers are likely to have more difficulty adjusting to eastward jet travel, early morning shift work and even regular work hours after a late weekend, as adjustment to these situations requires circadian phase advances.
The author would like to thank Dr Stephanie Crowley, Rose Diskin, Marissa Dziepak, Andrew Eiden, Sarah Garcia, Heather Gunn, Carlo Legasto, Thomas Molina, Jacqueline Munoz, Liz Sorkin, Christina Suh, Gabriela Velazquez and Nicole Woodrick for their assistance with data collection, Grace Padilla for scoring the sleep records, Dr Louis Fogg for his statistical advice, our Medical Director, Margaret Park, MD, and Dr Charmane Eastman for confirming the correct timing of the laboratory procedures. The author also thanks Dr Stephanie Crowley, Dr Mark Smith and Dr Charmane Eastman for their comments on the manuscript. Enviro-Med donated the light boxes. This project was supported by an R01 grant (HL083971) and administrative supplement (HL083971S1) from the National Heart Lung and Blood Institute to H. J. B. The content is solely the responsibility of the author, and does not necessarily represent the official views of the National Heart Lung and Blood Institute or the National Institutes of Health. The National Heart Lung and Blood Institute and the National Institutes of Health had no involvement in designing the study, data collection, data analysis and interpretation, writing of the manuscript, or in the decision to submit the manuscript for publication.