The estimated prevalence of DSPS in the present population-based study was 3.3%. The overlap between insomnia and DSPS was high, with half of the adolescents reaching the criteria for DSPS also presenting with symptoms according to the insomnia criteria. Adolescents with DSPS had a higher rate of school absence than their peers, with an independent contribution of DSPS after adjusting for socio-demographical factors, depression and insomnia. Girls had a higher prevalence of DSPS than boys, but the level of school absence was higher in boys with DSPS.
The current prevalence rate of DSPS is comparable to the study by Johnson et al. from 2006 . While both being community based studies, using similar operationalization, the adolescents in the present study were somewhat older, with an age spanning from 16 to 19 years compared to 13 to 16 years in the Johnson et al.’s study, which may contribute to explain the slightly higher prevalence rates in the current study. Not surprising, the current estimate is lower than reported in a previous Norwegian study from 2012 by Saxvig et al. (8.8%) which used less stringent criteria of delayed sleep phase . However, when restricting the definition in the abovementioned study to only include those with oversleeping two or more times per week, the prevalence was reduced to 4.9%, which is more in accordance with the rate reported in the present study.
The sleep pattern among adolescents with DSPS exhibited the expected sleep characteristics, with shorter sleep duration due to later bedtime and early awakenings during the weekdays, as well as rebound sleep during the weekends, while their WASO was not significantly different from that of their peers. Adolescents with DSPS also had lower sleep efficiency and higher sleep deficiency compared to peers without DSPS. The present study confirmed that the sleep pattern of adolescents with DSPS is located at an extreme end of a continuum of normal sleep. According to Gradisar et al. , it may be difficult to distinguish characteristics of DSPS from sleep patterns that are normal during adolescence, and that one therefore runs the risk of underestimating the prevalence of DSPS. However, just representing the extreme of a continuum may also lead to an overestimation of the diagnosis: Whereas the diagnostic criteria remain the same across the life span, the sleep characteristics fluctuate depending on the age. The misalignment criteria of the DSPS is at a level that might be regarded as the norm for adolescents, with the observed mean difference in bedtime and risetime for weekdays versus weekends being 2:15 hours and 4:28 hours respectively . What perhaps differentiate these normal sleep patterns from a diagnostic level of DSPS may therefore only partly be the sleep characteristics; more important are their functional impact and outcomes.
In accordance with the study by Johnson et al. , the present study defined oversleeping as one of the criteria of DSPS. Other functional sleep related factors such as tiredness and sleepiness was significantly higher in the DSPS group, but also prevalent in adolescents in general. The rate of frequent oversleeping was, however, not as prevalent, which in turn impacts the prevalence rate in the current study. This is also an example of how operationalization in epidemiological studies will impact the prevalence rates. We chose to use an operationalization with specific criteria that have been used previously. The new DSM-V diagnostic criteria on sleep wake disorder have chosen to use broad criteria to increase reliability in clinical diagnosis, making specific operationalizations for use in research more challenging. The current results still can inform on key aspects of the DSM-V diagnosis, including the importance of both insomnia as a co-occurring condition, and the functional impact on school attendance.
There was a significant association between DSPS and insomnia in the present sample, with half of the adolescents with DSPS also meeting the criteria for insomnia, as defined according to Lichstein’s Quantitative Criteria . The conclusion by Johnson et al.  that insomnia does not account for a significant proportion of DSPS was supported by the current study, as most of the adolescents with insomnia did not have DSPS. Thus, insomnia warrants attention as a public health concern in its own right also in adolescence, and should not be viewed as merely a byproduct of DSPS. While insomnia is frequent in adolescents with DSPS, and thus in line as one of the functional impairment criteria in DSM-V (and as such can be regarded as a symptom of DSPS), many proportion of adolescents with DSPS do not have insomnia, and the impact of having one or both of the conditions is still not settled. It may be that while the misaligned sleep schedule temporarily precedes the insomnia, the problems with maintaining sleep may be accompanied by worry and catastrophic thoughts that may exacerbate the sleep problems and warrant clinical attention in its own right.
Few population-based studies on adolescents have assessed potential gender differences in DSPS, most likely due to the low overall prevalence rates and the need of large samples to detect statistically significant gender differences. In the previous epidemiological study by Johnson et al. , no gender specific rates were reported. The female preponderance in the present study was surprising given the higher male ratio in the study by Thorpy et al. , although a more even gender balance was found in a broader delayed sleep phase study from Norway . Methodological differences may account for some of these differences. While girls had a higher rate of DSPS, the gender differences were not as marked as for insomnia, where there was an even higher female preponderance .
Adolescents with DSPS had a higher rate of non-attendance at school than their peers, emphasizing the functional impact of DSPS and mirroring the functional impact of sleep problems in adults. While girls had a higher rate of DSPS, the present study found a considerable higher odds of school absence in boys compared to girls. These results could illuminate some of the previously inconsistent gender prevalence rates. If the functional impact is higher among boys, as measured by an important outcome measure (such as school absence), this could instigate worry and health-seeking behavior among parents and teachers, and thus a diagnosis of DSPS would be more likely. This should lead to future research questions related to differences in characteristics of those who have the diagnosis of DSPS and those who meet all the diagnostic criteria except the functional impairment. It would also be of interest to examine if the same relations exist across other functional impairments, such as social relations and use of health care services. How school absence may impact academic performance remains unknown, although this may be one pathway linking DSPS to lower self-reported school grades found among adolescents with delayed sleep phase found in a previous study .
There are some methodological limitations of the present study that should be noted. First, the cross-sectional nature of the study does not allow for causal inferences. Thus, longitudinal studies concerning the relationship between pre-pubertal sleep patterns and later development of DSPS are needed to shed light on the developmental patterns. Secondly, tthe definition of DSPS represents another important limitation of the present study as it is based solely on self-report, and consequently lacks clinical evaluation and measurement by actigraphy or sleep diary. However, this is rarely possible in large epidemiological studies. Furthermore, while we did assess depression and insomnia, which accounted for some of the functional impact of DSPS in the present study, there may be other covariates not addressed in the current study that may explain parts of this association. For example, sleep phase misalignment may be a marker of more serious psychiatric disorders, which again may be related to school absence, such as a prodromal phases for more severe psychiatric disorders (e.g., bipolar disorder or schizophrenia). Third, depression was assessed by a self-report instrument, the SMFQ. As no validated cut-off exists for Norwegian adolescents, the 90th percentile on the total SFMQ score was chosen as an operationalization of depression. Clearly, this does not imply existence of a clinical diagnosis, such as MDD, and the lack of clinical interview in confirming a clinical diagnosis of depression is a limitation of the present study. In relation to this it should be noted that the SMFQ neither contains any sleep items nor items that assess any other vegetative symptoms. In contrast to conventional depression rating scales which normally contain such items, this prevents circularity and make the interpretation of associations between symptoms of sleep and affective problems unambiguous in the present study. Tiredness was included in the SMFQ, but the association to several sleep parameters was not higher for this item than for other depressive symptoms. Another limitation comprises the inclusion of a relatively low number of adolescents not in school. Although the few adolescents not attending school in the present study did not have a higher rate of DSPS, a higher participation rate among those adolescents would be needed to draw conclusions regarding this group specifically. Finally, the attrition from the study could affect generalizability, with a response rate of about 53% and with adolescents in schools overrepresented. Based on previous research from the former waves of the Bergen Child Study, non-participants often have more psychological problems than participants , and it is therefore likely that prevalences of both DSPS, insomnia and depression may be underestimated in the current study.
The high degree of overlap between DSPS and other conditions, such as depression and insomnia, warrants a thorough diagnostic evaluation and differential diagnosis when adolescents present with DSPS symptoms. While DSPS may preclude a diagnosis of insomnia, the symptomatic presentation may impact treatment choices.
The total score on the SMFQ in adolescents with DSPS in the current study (8.8) was comparable to the pre-treatment depression score in a RCT including adolescents with DSPS (total score SMFQ 7.5) . In that study, the treatments effect of a combined cognitive behavioral therapy and bright light therapy showed a large reduction of depression symptoms, suggesting that both the sleep problems and the co-occurring depressive symptoms may be targeted through the same interventions .