- Top of page
Delayed sleep phase disorder is characterized by a delay in the timing of the major sleep period relative to conventional norms. The sleep period itself has traditionally been described as normal. Nevertheless, it is possible that sleep regulatory mechanism disturbances associated with the disorder may affect sleep duration and/or architecture. Polysomnographic data that may shed light on the issue are scarce. Hence, the aim of this study was to examine polysomnographic measures of sleep in adolescents and young adults with delayed sleep phase disorder, and to compare findings to that of healthy controls. A second aim was to estimate dim light melatonin onset as a marker of circadian rhythm and to investigate the phase angle relationship (time interval) between dim light melatonin onset and the sleep period. Data from 54 adolescents and young adults were analysed, 35 diagnosed with delayed sleep phase disorder and 19 healthy controls. Results show delayed timing of sleep in participants with delayed sleep phase disorder, but once sleep was initiated no group differences in sleep parameters were observed. Dim light melatonin onset was delayed in participants with delayed sleep phase disorder, but no difference in phase angle was observed between the groups. In conclusion, both sleep and dim light melatonin onset were delayed in participants with delayed sleep phase disorder. The sleep period appeared to occur at the same circadian phase in both groups, and once sleep was initiated no differences in sleep parameters were observed.
- Top of page
Delayed sleep phase disorder (DSPD) is a circadian rhythm sleep disorder first described by Weitzman et al. (1981). It is characterized by a delay in the timing of sleep and often manifested as complaints of prolonged sleep onset latency and problems awakening in the morning (American Academy of Sleep Medicine, 2005). The disorder is particularly common among adolescents and young adults, with reported prevalence rates ranging from 0.5 to 16% (American Academy of Sleep Medicine, 2005; Gradisar et al., 2011; Hazama et al., 2008). We have recently reported a prevalence of delayed sleep phase of 8.4% in a large sample of Norwegian adolescents, with 68% reporting additional symptoms suggestive of DSPD (5.7% of the total sample) (Saxvig et al., 2012). Delayed sleep phase was associated with lower school grades, smoking, alcohol usage and elevated anxiety and depression scores (Saxvig et al., 2012), in agreement with previous findings of negative associations of DSPD (Crowley et al., 2007).
The pathophysiology of DSPD is still largely unknown (American Academy of Sleep Medicine, 2005). The delayed sleep period in patients with DSPD is assumed to reflect a delay of the endogenous circadian rhythm. Many studies have shown delayed markers of circadian rhythm, such as dim light melatonin onset (DLMO) (Chang et al., 2009; Shibui et al., 1999; Wyatt et al., 2006) and core body temperature nadir (Chang et al., 2009; Watanabe et al., 2003) in patients with DSPD. The endogenous phase delay may be caused by abnormally long spontaneous circadian periods (Regestein and Monk, 1995) or by a reduced ability to adequately phase advance (Czeisler et al., 1981; Ozaki et al., 1996). However, endogenous circadian rhythm disturbances may not be the sole cause underlying the disorder. Several reports describe elongated phase angle relationships (time interval) between measures of circadian rhythmicity and sleep, indicating that sleep may occur at a later circadian phase in patients with DSPD (Campbell and Murphy, 2007; Shibui et al., 1999; Watanabe et al., 2003), although neither Wyatt et al. (2006) nor Chang et al. (2009) found such differences. Uchiyama et al. (1999) have described reduced sleep propensity in patients with DSPD after sleep deprivation, indicating that the homeostatic process in DSPD differs from healthy controls. Alterations in homeostatic build-up and/or dissipation of sleep need may account for disturbed timing of sleep.
It has been assumed that sleep itself is normal in patients with DSPD, at least when conventional bedtimes and wake-up times are not enforced (American Academy of Sleep Medicine, 2005; Weitzman et al., 1981). This is despite the fact that the sleep regulatory mechanism disturbances responsible for the sleep phase delay also may influence the duration and/or architecture of sleep. If homeostatic processes are involved in the development or maintenance of DSPD, it should somehow be reflected in the amount or distribution of slow wave sleep (SWS). Moreover, as the circadian timing system is important for sleep duration as well as for the distribution of certain sleep stages (Dijk, 1999), it is plausible to assume that sleeping at a later endogenous time (longer phase angle) may have implications for both sleep duration and architecture.
Few studies have reported polysomnographic (PSG) measures of sleep in patients with DSPD, and the majority of existing reports does not include a control group (Alvarez et al., 1992; Rahman et al., 2009; Thorpy et al., 1988; Uchiyama et al., 1992; Weitzman et al., 1981). In fact, we are aware of only one such controlled study (Watanabe et al., 2003), and findings were somewhat contrary to the statements in the second edition of the International Classification of Sleep Disorders (American Academy of Sleep Medicine, 2005). The authors reported disturbed sleep including prolonged sleep onset latency (SOL), more wake after sleep onset (WASO), reduced sleep efficiency (SE) and reduced SWS in 11 patients with DSPD compared to 11 healthy controls. The reduced sleep quality was attributed to a longer phase angle in which sleep offset was delayed in relation to the core body temperature nadir (Watanabe et al., 2003). Campbell and Murphy (2007) recorded PSG and body temperature in one patient with DSPD and three controls in temporal isolation. Although not generalizable, these findings were similar to those reported by Watanabe et al. (2003) in terms of disturbed sleep, altered distribution of SWS and altered phase angle relationship.
Hence, the assumption that the sleep period itself is normal in patients with DSPD when allowed to sleep at preferred times has not been supported by scientific evidence. Studies are warranted to explore sleep duration and architecture in patients with DSPD and its relation to the endogenous circadian rhythm when in isolation from external constraints. Such knowledge will have implications for our understanding of pathophysiology, diagnostics and treatment. Thus, the primary objective of the present study was to investigate polysomnographic measures of sleep in adolescents and young adults with DSPD and compare findings to those of healthy controls. A second aim was to estimate circadian phase (DLMO) and to explore phase-angle relationships in relation to the sleep period. Sleep was assessed on a self-chosen sleep/wake schedule, to rule out effects of external constraints.
- Top of page
The sleep period was delayed in participants with DSPD compared to the controls. As shown in Table 2, this includes later times for lights off, sleep onset, sleep offset and lights on. Although participants with DSPD spent more time in bed, TST was not significantly different between the DSPD and the control group. The higher TWT in the DSPD group was attributed mainly to longer SOL. Both WASO and EMA were similar in the two groups. SE was not statistically different between the groups. Once sleep onset had occurred, no group differences in sleep architecture were observed. Time spent in each sleep stage during the sleep period did not differ between the groups (Table 2). Furthermore, sleep stage distribution in the first, middle and last third of the sleep period did not differ between the groups (Table 3). Similar proportions of SWS were observed in the two groups both during the first 3 h of sleep (48.8 ± 10.5% in the DSPD group compared to 48.1 ± 7.7% in the control group, P = 0.81) and during the last 3 h of sleep 11.8 ± 7.5% in the DSPD group compared to 12.0 ± 9.9% in the control group, P = 0.92). However, participants with DSPD had more SWS in the time interval between 06:00 and 08:00 hours, the time at which most participants have to wake up for school or work (24.4 ± 18.0% in the DSPD group, 9.7 ± 9.6% in the control group, P < 0.0005).
Table 2. Sleep parameters derived from polysomnographic recordings in participants with DSPD and controls
| ||Controls (n = 19)||DSPD (n = 35)||Cohen's d|
|Lights off||23:50 ± 69 min||02:32 ± 124 min**||−1.61|
|Sleep onset||00:07 ± 75 min||03:08 ± 124 min **||−1.77|
|Sleep offset||08:55 ± 110 min||12:44 ± 117 min**||−2.02|
|Lights on||09:01 ± 70 min||12:55 ± 113 min**||−2.49|
|TIB (min)||551 ± 67.2||622 ± 112.2*||−0.77|
|TST (min)||507 ± 68.8||549 ± 116.5||−0.44|
|SOL (min)||17 ± 12.6||36 ± 49.7*||−0.52|
|WASO (min)||21 ± 18.1||26 ± 32.8||−0.17|
|EMA (min)||6 ± 7.6||11 ± 15.3||−0.41|
|TWT (min)||44 ± 24.2||73 ± 59.7*||−0.64|
|SE (%)||92 ± 4.7||88 ± 10.1||0.51|
|N1 (%)||9 ± 3.0||9 ± 3.8||0.00|
|N2 (%)||40 ± 5.6||41 ± 6.6||−0.16|
|N3 (%)||26 ± 6.0||26 ± 6.3||0.00|
|REM sleep (%)||24 ± 3.0||24 ± 4.3||0.00|
|REM sleep latency (min)||96 ± 48.5||80 ± 37.8||0.37|
Table 3. Distribution of wake and sleep stages during the first, middle and last third of the sleep period in participants with delayed sleep phase disorder (DSPD) and controls
| ||Controls (n = 19)||DSPD (n = 35)||Cohen's d|
|Wake time (%)|
|First third||3.0 ± 3.9||6.4 ± 13.7||−0.33|
|Middle third||4.0 ± 5.3||3.8 ± 3.9||0.04|
|Last third||4.7 ± 5.0||3.5 ± 2.6||0.30|
|First third||7.3 ± 3.1||7.2 ± 3.5||0.03|
|Middle third||8.1 ± 4.4||8.3 ± 5.4||−0.04|
|Last third||11.0 ± 4.9||11.2 ± 4.9||−0.04|
|First third||28.4 ± 7.5||29.6 ± 9.4||−0.14|
|Middle third||46.6 ± 10.1||43.3 ± 10.2||0.33|
|Last third||40.9 ± 8.8||43.2 ± 9.0||−0.26|
|First third||46.7 ± 8.0||43.6 ± 11.2||0.32|
|Middle third||17.9 ± 15.9||18.6 ± 10.8||−0.05|
|Last third||11.5 ± 7.5||11.4 ± 6.8||0.01|
|REM sleep (%)|
|First third||14.7 ± 8.9||13.2 ± 8.6||0.17|
|Middle third||23.4 ± 8.5||26.0 ± 8.2||−0.31|
|Last third||31.8 ± 8.1||30.6 ± 6.7||0.16|
Early evening concentrations of melatonin appeared similar in the two groups, as seen in Fig. 1, where average saliva melatonin concentrations at each sampling time between 19:00 and 02:00 hours are displayed.
Figure 1. Average saliva melatonin concentration in delayed sleep phase disorder (DSPD) participants and controls in the time interval between 19:00 and 02:00 hours. Values are included only when at least three samples are collected at a given time. Positive standard deviations are indicated. *P < 0.05, independent t-test.
Download figure to PowerPoint
Average DLMO in the DSPD group was delayed by more than 3 h compared to the control group (Table 4). However, the sleep period appeared to occur at the same circadian phase in both groups, as phase angle relationships did not differ significantly between the two groups, neither with respect to time for sleep onset nor offset (Table 4). In many participants no rise in melatonin concentration was observed during the sampling period, and DLMO could be calculated in only a subsample of the participants, 20 in the DSPD group (57.1%) and eight controls (42.1%). The times for collection of the last saliva sample in participants where DLMO could be calculated are presented in Table 4. The corresponding times for participants where no rise in melatonin concentration was observed were 22:27 ± 31 min in the control group (n = 11, P = 0.308 compared to the controls where DLMO could be calculated) and 00:12 ± 99 min in the DSPD group (n = 20, P = 0.080 compared to the DSPD where DLMO could be calculated, P = 0.001, compared to the controls where no rise in melatonin concentration was observed). In the participants where no rise in melatonin concentration was observed, the last saliva sample was collected on average 138.7 ± 81.5 min earlier than their bedtime on the previous night (PSG night) compared to 46.4 ± 107.2 min (P = 0.001) in the participants where DLMO could be calculated. The times for collection of the last saliva samples for all participants were 22:35 ± 36 min in the control group and 00:53 ± 121 min in the DSPD group, respectively (P < 0.0005).
Table 4. Dim light melatonin onset (DLMO), time for collection of the last saliva sample and phase angle with respect to sleep onset and offset in participants with DSPD and controls
| ||Controls (n = 8)||DSPD (n = 20)||Cohen's d|
|DLMO||21:08 ± 73 min||00:15 ± 133 min*||−1.74|
|Last saliva sample||22:45 ± 42 min||01:24 ± 128 min*||−1.67|
|Phase anglesleep onset||2 h 47 ± 126 min||2 h 17 ± 121 min||0.24|
|Phase anglesleep offset||11 h 23 ± 90 min||12 h 08 ± 81 min||−0.53|
- Top of page
The main objective of this study was to compare PSG measures of sleep in adolescents and young adults with DSPD to that of healthy controls. Participants were instructed to follow a self-chosen sleep schedule for 3 days prior to PSG recording. The protocol differs as such from protocols where participants go to bed at a fixed time (e.g. Rahman et al., 2009), and yields new information about sleep in DSPD in absence of social restraints.
Our results show that the major differences between the two groups were related to the timing of sleep, confirming the diagnosis. Prolonged SOL was observed in the DSPD group, even though participants went to bed at a self-chosen time. The sleep onset problems associated with DSPD have commonly been attributed to attempts of going to sleep at circadian phases not optimal for sleep (American Academy of Sleep Medicine, 2005). Although a SOL of 37 min, such as found in the present study, is only slightly longer than what is considered normal (30 min or less; Pallesen et al., 2008), it is possible that previous experiences of not being able to fall asleep have elicited a conditioned problem with sleep onset (Lack and Wright, 2007).
Although mean TST differed by 40 min between the groups, this difference did not reach statistical significance due to large interindividual differences, in particular in the DSPD group. SE, the amount of the different sleep stages (REM sleep, N1, N2, N3) and overall distribution of the sleep stages did not differ between the groups, as seen by similar amounts of W, N1, N2, N3 and REM sleep in the first, middle and last third of the night, respectively, as well as similar amounts of SWS in the two groups both during the first 3 h and the last 3 h of sleep. Hence, our findings are in line with the statement in the International Classification of Sleep Disorders of normal polysomnograms in patients with DSPD when performed at preferred sleep times (American Academy of Sleep Medicine, 2005).
A core symptom of DSPD, in addition to long SOL, is difficulty in waking up at socially acceptable times. Our analyses show that the participants with DSPD have considerable amounts of SWS between 06:00 and 08:00 hours, the time when most have to get up for school or work. This finding is a logical consequence of the sleep phase delay, as the amount of SWS usually diminishes as sleep progresses, but is illustrative of the challenges faced by the patients. Arousal threshold is higher from SWS than from lighter sleep stages, making it harder for patients with DSPD to wake up at these early hours. Moreover, awakenings from SWS are associated with more sleep inertia/drunkenness, which can impair performance (Matchock, 2010).
A second aim of the study was to estimate DLMO as a marker of circadian phase. DLMO could be calculated in only a subsample of the participants (20 DSPD and eight controls). In all but two participants, saliva samples collected on the following morning (07:00 and/or 08:00 hours) showed high melatonin concentrations (≥4 pg ml−1). In one control all samples showed melatonin concentrations <4 pg ml−1, whereas one participant with DSPD failed to deliver adequate amounts of saliva for the morning samples. The increased morning melatonin concentrations argue against the possibility that participants were low melatonin secretors. It is possible, however, that behavioural factors suppressed evening melatonin temporarily. In the present study participants received no specific instructions regarding posture or physical activity. Occasional use of non-steroidal anti-inflammatory drugs was not prohibited. Both non-steroidal anti-inflammatory drugs and physical activity may suppress melatonin secretion. Because sampling was home-based, it was not possible to confirm that all instructions were followed. Nevertheless, home-based DLMO assessment appears to correlate well with laboratory assessment (Pullman et al., 2012), although it does not work in all patients due probably to differences in behaviour or environment, e.g. posture or light exposure (Pullman et al., 2012). In the subgroup of participants, where no rise in melatonin concentration was observed, the last saliva sample was collected more than 2 h earlier than their bedtime on the previous night (PSG night) compared to less than an hour in the subgroup where DLMO could be calculated. Thus, it appears that several participants may have gone to bed earlier than usual and consequently before their initial rise in melatonin concentration. This is due probably to a combination of factors related to the treatment study into which the DSPD participants were enrolled, such as self-chosen bedtime and wake-up time at 07:00 hours the following morning. Also, the sunglasses worn by the participants were very dark and probably limited the number of activities in which participants could engage.
Results from the subgroup in which DLMO could be calculated showed that participants in the DSPD group had DLMO more than 3 h later than participants in the control group, indicating a substantially delayed endogenous rhythm. Averaged DLMOs were, in both groups, similar to that found by Chang et al. (2009), but should be interpreted with caution due to the low sample size in which DLMO could be calculated in the present study. To take into account information obtained about melatonin secretion in cases where DLMO could not be calculated, times for the last saliva sample collected were compared between the groups. Also, these calculations indicated a circadian delay in the DSPD group, as the average time for the last sample collected occurred more than 2 h later in the DSPD group than in the control group. Phase angle with respect to sleep onset and offset differed between the groups by 30 and 45 min, respectively, but these differences did not reach statistical significance. These findings are consistent with results published by Wyatt et al. (2006) and Chang et al. (2009).
The results from the present study are in line with descriptions of DSPD in the International Classification of Sleep Disorders (American Academy of Sleep Medicine, 2005), and may support the contention by Chang et al. (2009) that when allowed to sleep on a schedule free from social restraints, sleep in patients with DSPD takes place at a normal circadian phase, allowing normal sleep quality. The results are, however, contrary to findings in the only previous controlled PSG study performed on patients with DSPD (Watanabe et al., 2003). These authors observed both reduced sleep quality and altered temporal distribution of SWS, which they attributed to an alteration in phase angle relationship. Slight differences in protocols may be the cause of these discrepancies. In the habitual sleep times protocol used by Watanabe et al. (2003), it appears that patients were instructed to go to bed at their usual time. Usual bedtimes may be influenced by external factors such as work, school or input from parents and partners, and the long phase angle may thus reflect an influence of external restraints on habitual sleep time rather than underlying pathophysiology. Moreover, the control group in that study slept at a fixed time (23:00–07:00 hours), whereas both groups in the present study received instructions to self-choose their bedtime. Our results are also in contrast to the findings of Campbell and Murphy (2007), where PSG and body temperature were recorded from a patient with DSPD in temporal isolation, hence controlling for external factors. That study is, however, limited by the low number of participants (one DSPD, three controls). It is worth mentioning that somewhat large descriptive differences in several parameters (e.g. TST and phase angle) failed to reach statistical significance due to large interindividual differences in the present study. It is possible that DSPD may arise from different pathophysiological sources, and that larger-scale studies may be needed to detect and to differentiate between them.
In terms of sample size the present study is, to our knowledge, the largest controlled study assessing objective measures of sleep in patients with DSPD. The protocol allowed participants to sleep until spontaneous awakening for 3 days prior to the PSG night used for data analyses and 4 days prior to saliva sampling, hence ruling out possible effects of chronic sleep debt or enforced asynchrony. PSG was performed ambulatory following an adaptation night, minimizing potential effects of the procedure on sleep. Saliva samples for DLMO assessment were also home-based. Participants were screened thoroughly for other pathologies; as such, the present study was well suited to address the issue of sleep duration and architecture in patients with DSPD and its relation to phase angle.
The study has some limitations, however. Participants in the DSPD group had volunteered to participate in a treatment study, causing a potential selection bias towards help-seeking patients. In estimating phase angle relationship, PSG was obtained on the night prior to DLMO sampling (nights 4 and 5, respectively). Hence, there is no direct temporal relationship between DLMO and sleep onset/offset parameters. However, it is not likely that DLMO changes greatly over such a short time interval (1 day), considering the findings by Wyatt et al. (2006) of temporal stability of DLMO. Previous reports on phase angle have often utilized averaged times for sleep onset as recorded by sleep diary or actigraphy. Sleep onset as found by PSG is presumably more accurate, but it does not confirm sleep patterns over time. In the present study, no sleep diary or actigraphy monitoring were utilized in association with PSG to confirm habitual sleep–wake schedules. One limitation regarding DLMO reflects the somewhat smaller sample in which DLMO could be calculated, as discussed above.
In conclusion, our results showed that both sleep and DLMO were delayed in adolescents and young adults with DSPD. Sleep architecture did not differ from that of the control group. Accordingly, the results support the statement in the International Classification of Sleep Disorders (2005) that the sleep period itself is normal in patients with DSPD. Moreover, although DLMO could be calculated in only about half the sample, the sleep period appeared to occur at the same circadian phase in both groups. Hence, it may seem that when in isolation from external restraints, patients with DSPD sleep at a normal circadian phase allowing normal sleep architecture.