Sleep patterns and insomnia among adolescents: a population-based study

Authors

  • Mari Hysing,

    Corresponding author
    1. Department of Biological and Medical Psychology, University of Bergen, Bergen, Norway
    • The Regional Centre for Child and Youth Mental Health and Child Welfare, Uni Health, Uni Research, Bergen, Norway
    Search for more papers by this author
  • Ståle Pallesen,

    1. Department of Psychosocial Science, University of Bergen, Bergen, Norway
    2. Norwegian Competence Center for Sleep Disorders, Haukeland University Hospital, Bergen, Norway
    Search for more papers by this author
  • Kjell M. Stormark,

    1. The Regional Centre for Child and Youth Mental Health and Child Welfare, Uni Health, Uni Research, Bergen, Norway
    2. Department of Clinical Psychology, University of Bergen, Bergen, Norway
    Search for more papers by this author
  • Astri J. Lundervold,

    1. The Regional Centre for Child and Youth Mental Health and Child Welfare, Uni Health, Uni Research, Bergen, Norway
    2. Department of Biological and Medical Psychology, University of Bergen, Bergen, Norway
    3. K. G. Jebsen Centre for Research on Neuropsychiatric Disorders, University of Bergen, Bergen, Norway
    Search for more papers by this author
  • Børge Sivertsen

    1. Division of Mental Health, Norwegian Institute of Public Health, Bergen, Norway
    2. Department of Clinical Psychology, University of Bergen, Bergen, Norway
    Search for more papers by this author

Correspondence

Dr Mari Hysing, Centre for Child and Adolescent Mental Health and Welfare, Postbox 7810, 5020 Bergen, Norway.

Tel.: +55-58-86-98;

fax: +47-55-58-98-78;

e-mail: mari.hysing@uni.no

Summary

The aim of the current study was to examine sleep patterns and rates of insomnia in a population-based study of adolescents aged 16–19 years. Gender differences in sleep patterns and insomnia, as well as a comparison of insomnia rates according to DSM-IV, DSM-V and quantitative criteria for insomnia (Behav. Res. Ther., 41, 2003, 427), were explored. We used a large population-based study in Hordaland county in Norway, conducted in 2012. The sample included 10 220 adolescents aged 16–18 years (54% girls). Self-reported sleep measurements included bedtime, rise time, time in bed, sleep duration, sleep efficiency, sleep onset latency, wake after sleep onset, rate and frequency and duration of difficulties initiating and maintaining sleep and rate and frequency of tiredness and sleepiness. The adolescents reported short sleep duration on weekdays (mean 6:25 hours), resulting in a sleep deficiency of about 2 h. A majority of the adolescents (65%) reported sleep onset latency exceeding 30 min. Girls reported longer sleep onset latency and a higher rate of insomnia than boys, while boys reported later bedtimes and a larger weekday–weekend discrepancy on several sleep parameters. Insomnia prevalence rates ranged from a total prevalence of 23.8 (DSM-IV criteria), 18.5 (DSM-V criteria) and 13.6% (quantitative criteria for insomnia). We conclude that short sleep duration, long sleep onset latency and insomnia were prevalent in adolescents. This warrants attention as a public health concern in this age group.

Introduction

During adolescence a range of biological, psychological and social factors interact, resulting in shortened sleep duration, in what has been characterized as ‘the perfect storm’ (Carskadon, 2011). Secular trends suggest that sleep deficiency and sleep problems are increasing among adolescents (Matricciani et al., 2012; Pallesen et al., 2008). As this shortened sleep duration is not accompanied by a reduction in sleep need during adolescence, a large proportion of adolescents experience sleep deficiency or insufficient sleep (Fallone et al., 2002), with possible negative consequences in terms of reduced daytime functioning and school performance (for a review, see Dewald et al., 2010), increased risk of mental health problems (Cousins et al., 2011) as well as increased risk of traffic accidents (Danner and Phillips, 2008).

Insufficient sleep has been defined as the duration of sleep below which waking deficits begin to accumulate (van Dongen et al., 2003), whereas sleep need has been defined as habitual sleep duration in the absence of pre-existing sleep debt (Dement and Grenber, 1966). Most studies rely solely upon sleep duration as an indicator of insufficient sleep, which may be inaccurate due to large variations in individual sleep need (Mercer et al., 1998). Others have defined insufficient sleep as sleep duration far below age-expected norms (Pallesen et al., 2011) or ask explicitly whether and to what degree the respondent has not obtained enough sleep (Altman et al., 2012). Perhaps a better approach to making it possible to express the magnitude of insufficient sleep is to calculate the discrepancy between self-reported sleep and perceived sleep need. This approach has also been used in previous studies (Hublin et al., 2001).

Adolescent sleep is characterized by a large discrepancy between weekdays and weekend sleep patterns, including a sleep phase shift to later bedtimes as well as an average of 1 and 2 h longer sleep durations during weekends (Crowley et al., 2007; Gradisar et al., 2011a). While adolescent sleep phase delay is well documented (Taylor et al., 2005), less is known about the time it takes adolescents to fall asleep, as assessment of sleep onset latency (SOL) is seldom included in general population-based studies. In a recent review of studies on sleep patterns among adolescents, only three single studies assessed SOL. These showed that between 20 and 26% of the samples took more than 30 min to fall asleep (Gradisar et al., 2011b). An Icelandic study reported an average SOL of 16.8 min among adolescents (Thorleifsdottir et al., 2002). The authors proposed that the definition of an acceptable/normal SOL may differ between adults and adolescents. This issue has not been resolved, and inclusion of SOL as a central sleep parameter in future studies of sleep among adolescents is therefore recommended (Gradisar et al., 2011b).

Prolonged SOL is also a defining characteristic of insomnia. The exact rate of insomnia among adolescents is uncertain (Roane and Taylor, 2008) due to large variations in operationalization across studies, thus complicating comparisons. In a European general population study of adolescents aged between 15 and 18 years, a 4% prevalence rate of insomnia according to the DSM-IV criteria was found (Ohayon et al., 2000). An American population-based study reported a 10.7% lifetime prevalence of insomnia according to the DSM-IV criteria, including a frequency criterion of 4 days a week (Johnson et al., 2006). The proposed revision in the DSM-V concerning insomnia disorder will probably affect the estimated prevalence, as it adds a minimum frequency criterion of 3 days to the diagnosis and increases the minimal duration threshold from 1 to 3 months. Similarly, estimated prevalence rates are different when using the suggested quantitative criteria for insomnia, recommending an operationalization of insomnia based on a thorough review of the literature in order to identify the most valid criteria (Lichstein et al., 2003). These criteria specify a SOL of more than 30 min, and insomnia occurring on three or more nights a week for at least 6 months. The prevalence of insomnia using the new proposed DSM-V definition or the quantitative criteria has, to our knowledge, not been assessed previously in population samples of adolescents.

While no gender differences in prepubertal children in insomnia rates were found, a more than twofold risk for insomnia was found in postmenes girls compared to boys in an American population-based study using the DSM-IV diagnostic criteria (Johnson et al., 2006). However, there are few studies regarding gender differences in insomnia in adolescence, and potential gender differences based on DSM-V and research criteria for insomnia in adolescence have, to the best of our knowledge, not been investigated.

Based on the above considerations, the aim of the present study was to characterize sleep patterns in adolescence, including gender differences, in a large population-based study. The second aim was to assess the rate of insomnia according to the definitions of DSM-IV, DSM-V and quantitative criteria for insomnia, analysed separately for girls and boys.

Methods

In this population-based study, we employed information from the ung@hordaland survey of adolescents in the county of Hordaland in western Norway. All adolescents born between 1993 and 1995 and all students attending secondary education during spring 2012 were invited to participate in the ung@hordaland survey, the main aim of which was to assess mental health problems and service use in adolescents, with a special emphasis on the prevalence of mental health problems. The data were collected during spring 2012. Adolescents in upper secondary education received information via e-mail, and one school hour was allocated for them to complete the questionnaire at school. Those not at school received information by postal mail to their home addresses. The questionnaire was web-based, and covered a broad range of mental health issues, daily life functioning, use of health care and social services, demographic background variables and a request for permission to obtain school data, and to link the information with national health registries and parental questionnaires. Uni Health collaborated with Hordaland County Council in conducting of the study. The study was approved by the Regional Committee for Medical and Health Research Ethics in western Norway. The current study is based on the first version of data files released in May 2012.

Sample

All adolescents born between 1993 and 1995 (= 19 430) were invited to participate in the current study, which took place during the first months of 2012, 10 220 of whom agreed, yielding a participation rate of 53%. Sleep variables were checked for validity of answers based on preliminary data analysis, resulting in 374 subjects being omitted due to obvious invalid responses (e.g. negative sleep duration and sleep efficiency). Thus, the total sample size in the current study was 9875.

Instruments

Demographic information

All participants indicated their vocational status, with response options being ‘high school student’, ‘vocational training’ or ‘not in school’. Maternal and paternal education were reported separately, with four response options: ‘primary school’, ‘secondary school’, ‘college or university: less than 4 years’ and ‘college or university: 4 years or more’.

Sleep variables

Self-reported bedtime and rise time were reported separately for weekends and weekdays. Time in bed (TIB) was calculated by subtracting bedtime from rise time. SOL and WASO were reported in hours and minutes, and sleep duration was defined as TIB minus SOL and WASO. For the purpose of the present study, sleep duration was also split into ten categories (<4, 4–5, 5–6, 6–7, 7–8, 8–9, 9–10, 10–11, 11–12 and >12 h). Sleep efficiency was calculated as sleep duration divided by TIB multiplied by 100 (reported as a percentage). Subjective sleep need was reported in hours and minutes, and sleep deficiency was calculated separately for weekends and weekdays, subtracting total sleep duration from subjective sleep need.

Difficulties initiating and maintaining sleep (DIMS) were rated on a three-point Likert scale, with response options of ‘not true’, ‘somewhat true’ and ‘certainly true’. Given a positive response (‘somewhat true’ or ‘certainly true’), the participants were then asked how many days per week they experienced problems either initiating or maintaining sleep. The participants also provided information on the duration of DIMS.

A joint question on tiredness/sleepiness was rated on a three-point Likert-scale with response options of ‘not true’, ‘somewhat true’ and ‘certainly true’. If confirmed (‘somewhat true’ or ‘certainly true’), participants reported the number of days per week on which they experienced sleepiness and tiredness, respectively.

Three operationalizations of insomnia were investigated (see Table 1). All three definitions included a positive response (‘somewhat true’ or ‘certainly true’) to (i) DIMS and (ii) sleepiness and/or tiredness (see above). In addition, to fulfil the criteria for DSM-IV insomnia, duration of DIMS for at least 1 month was also required. For DSM-V insomnia, the following additional criteria were required: a DIMS frequency of at least 3 days per week and duration of insomnia of at least 3 months. The quantitative criteria for insomnia (Lichstein et al., 2003) were operationalized as follows: reporting DIMS at least three times a week, with a duration of 6 months or more, in addition to reporting SOL and/or WASO of more than 30 min.

Table 1. Overview of the insomnia diagnosis in DSM-IV, the proposed revision in DSM-V and the quantitative criteria for insomnia
 Duration of insomniaFrequency of insomniaSOL and WASODaytime functioning
  1. SOL, sleep onset latency; WASO, wake after sleep onset.

Diagnostic criteria for insomnia
DSM-IV1 monthNoneNot specifiedNot specified
DSM-V3 months3 nights/ weekNot specifiedX
Quantitative criteria>6 months3 nights/week≥31 minX

Statistics

IBM spss version 20 (SPSS Inc., Chicago, IL USA) for Windows was used for all analyses. Multivariate analysis of variance (manova) was used to examine gender and age differences on the sleep variables, as well as interaction effects between age and gender. Chi-square tests were used to examine differences in sleep duration (10 different duration categories) between weekdays and weekends. Gender differences in insomnia prevalence were estimated using logistic regression analyses using gender as the exposure variables and the three insomnia definitions (DSM-IV insomnia, DSM-V insomnia and quantitative criteria for insomnia) as outcome variables. To investigate whether gender differences were significantly different across the three operationalizations, we also calculated the relative risk ratio (RRR), as recommended by Altman and Bland (2003), in order to test for significant differences between the odds ratios. This is a well-established test of interaction to compare estimates on a log scale.

Results

Of the adolescents born between 1993 and 1995 (with a mean age of 17 years), 53.5% of the participants were girls; the majority comprised high school students (98%). For details and information on socioeconomic status and demographic information, see Table 2.

Table 2. Demographical variables in the ung@hordaland study (= 9846)
 MeanSD
Age (years)17.00.87
 % n
  1. SD, standard deviation.

Girls53.55215
Vocational situation
High school student97.99219
Vocational training1.4132
Not in school0.767
Maternal education
Primary school7.7742
Secondary school31.43042
College/university (<4 years)15.21469
College/university (4+ years)21.82112
Paternal education
Primary school7.9763
Secondary school34.73343
College/university (<4 years)9.5920
College/university (4+ years)22.72186

Sleep patterns

Bedtime, rise time, TIB and sleep duration for the total sample, stratified by gender, are presented in Table 3. The mean bedtime on weekdays was 23:18 hours, significantly later for boys (23:56) than girls (23:10). Mean TIB for weekdays was 7:29 hours (boys: 7:26 and girls: 7:32), whereas mean sleep duration was 6:25 hours (boys: 6:28 and girls: 6:22).

Table 3. Sleep characteristics in the ung@hordaland study (= 9846)
 GirlsBoysP-valueTotal
MeanSDMeanSDMeanSD
  1. P-level indicates significant differences between girls and boys. SD, standard deviation.

Weekdays
Bedtime23:100:5723:261:01<0.00123:180:59
Rise time6:420:406:530:41<0.0016:470:41
Time in bed7:321:017:261:02<0.0017:291:01
Sleep duration6:221:406:281:370.0066:251:39
Sleep efficiency (%)84.218.086.617.2<0.00185.3%17.7
Weekends
Bedtime1:171:221:521:34<0.0011:331:29
Rise time11:061:2411:261:38<0.00111:151:31
Time in bed9:481:199:341:26<0.0019:411:23
Sleep duration8:391:508:351:520.0188:371:51
Sleep efficiency (%)88.013.889.613.5<0.00188.7%13.7
Weekdays/weekends
Sleep onset latency0:510:580:430:56<0.0010:470:57
Wake after sleep onset0:170:380:120:40<0.0010:150:39
Subjective sleep need8:431:438:262:03<0.0018:351:54

Bedtime during weekends was, on average, 2 h and 25 min later than on weekdays (01:13), while the corresponding rise time discrepancy was 4 h and 28 min, reflecting that the adolescents slept on average 2 h and 12 min more during weekends than on weekdays. Both rise- and bedtime discrepancies between weekdays and weekends were significantly larger for boys than girls (< 0.001). Distribution of sleep duration on weekdays and at weekends are presented in Fig. 1.

Figure 1.

Sleep duration on weekdays and at weekends. Error bars represent 95% confidence intervals.

The adolescents' subjective sleep need was 8 h and 35 min, yielding a sleep deficiency on weekdays of 2 h and 9 min. No sleep deficiency was found for weekends.

SOL, WASO and sleep efficiency

Mean SOL was 47 min, with 24.2% reporting SOL less than 15 min and 59% reporting SOL longer than 30 min (see Fig. 2 for details). SOL was significantly longer for girls than boys (< 0.001). Mean WASO was 15 min, and 79% of the adolescents reported less than 15 min WASO. Mean sleep efficiency during weekdays was 85%, with girls (84%) having lower sleep efficiency than boys (87%). Higher sleep efficiency was observed at the weekend, but with similar gender differences (88% versus 90% for girls and boys, respectively (< 0.001).

Figure 2.

Sleep onset latency (SOL) and wake after sleep onset (WASO) among adolescents in the ung@hordaland study (= 9846).

Insomnia

The prevalence of insomnia was calculated for the total sample and separately for the two gender groups using three different definitions. All insomnia definitions included an algorithm of difficulties initiating and/or maintaining sleep and tiredness and/or sleepiness during daytime. The prevalence estimates ranged from a total prevalence of 23.8%, using the DSM-IV criteria, to 18.5% according to the proposed DSM-V criteria, expanding the duration from 1 to 3 months and including frequency criteria of 3 days per week (See Fig. 3). According to the quantitative criteria for insomnia, the prevalence was 13.6%. Girls had a significantly higher prevalence of insomnia across all three insomnia definitions, but there were no significant differences in the magnitude of gender differences between the three diagnostic criteria, as calculated by the RRR (all 95% confidence intervals included the value 1.0).

Figure 3.

Prevalence of insomnia according to different operationalizations, stratified by gender. Error bars represent 95% confidence intervals.

Age, gender and interaction effects

Significant gender differences were found for most sleep variables, as detailed in Table 3. There was a significant age effect on some sleep variables, with the youngest age cohorts reporting earlier bedtimes and rise times both on weekdays and at weekends, and also sleeping longer at weekends. The youngest cohort also spent more TIB, had larger subjective sleep need, had more sleep deficiency during weekdays and reported less insomnia than older adolescents (all Ps < 0.001). There were no significant age differences in SOL, WASO and sleep efficiency.

Few significant interaction effects between age and gender were observed, except for bedtime, rise time and TIB during weekdays (all Ps < 0.001). For example, while TIB increased with age for girls, a corresponding decrease with age was observed for boys (< 0.001).

Discussion

To sum up the main findings: sleep patterns of adolescents between 16 and 19 years were characterized by late bedtimes, long SOL and a short sleep duration, contributing to a daily sleep deficiency of about 2 h on weekdays. A high rate of insomnia was evident across the diagnostic definitions, with total prevalence ranging from 23.8% using the DSM-IV criteria to 18.5% according to the proposed DSM-V criteria and to 13.6% using the quantitative criteria for insomnia. Girls had a significantly higher prevalence of insomnia than boys across all three insomnia definitions.

Information about sleep patterns during weekdays revealed late bedtimes. Rise time on school days showed limited variations, due probably to the fact that most of the adolescents were high school students with fixed and early school starting times. To obtain the recommended sleep duration of 8–9 h, which was also in accordance with their self-perceived sleep need, they should have gone to bed at around 22:00 hours. At weekends, their sleep duration was in accordance with their subjective sleep need, with a shift towards later bedtime of more than 2 h and an even later rise time. The adolescent's sleep pattern in the present study confirmed sleep phase delays and the late bedtimes reported in previous studies of adolescents (Crowley et al., 2007; Gradisar et al., 2011b).

The mean sleep duration of approximately 6½ h on weekdays shown in the present study is, however, shorter than that reported in most previous studies. One reason could be differences in definitions, methods and samples between our and other studies of adolescent sleep duration. For example, in the Ohayon study the mean sleep duration in a European sample of 15–18-year-olds was approximately 8 h (Ohayon et al., 2000). This was methodologically comparable to the present study, as both calculated sleep duration by subtracting sleep latency from bedtime. Other studies have used more dissimilar definitions. We have, for example, defined sleep duration by subtracting reported wake time (SOL and WASO), whereas results from many previous studies are based on TIB only. In their review of international sleep studies, Gardiner et al. suggested that the reported sleep duration was probably overestimated by about a quarter of an hour. The review of international sleep studies (Gradisar et al., 2011b) suggested that the sleep duration reported in their overview was probably overestimated, as it was calculated based on the SOL in an icelandic study that reported SOL of about 15 minutes (Thorleifsdottir, et al., 2002). Mean SOL was 47 min in the present study, which is more than twice as much as reported previously in studies of SOL in adolescents (Gradisar et al., 2011b), and a considerably longer mean SOL than reported in the aforementioned Icelandic study (Thorleifsdottir et al., 2002). The fact that the SOL was longer in the present study than in most previous studies may reflect cohort effects. This interpretation is in line with studies showing a secular increase in the proportion of adolescents reporting sleep onset difficulties (Pallesen et al., 2008).

Another possible reason for the observed discrepancies between the findings in the current and previous studies may be seasonal variations in terms of daylight illumination during the time of data collection. The data in the current study were collected from February to May 2012, and while we cannot rule out a possible seasonal influence on our data, a recent study from Norway investigating seasonal variation in insomnia and sleep duration found no evidence of such a seasonal effect (Sivertsen et al., 2011).

The adolescents reported a mean sleep need of between 8 and 9 h. While there are individual differences, the mean reported sleep need in the present study is in accordance with the empirically derived suggested sleep need in adolescents of about 8–9 h (Carskadon et al., 1980). In the present study, their sleep duration is in accord with their subjective sleep need during weekends, as they probably do not have such specific demands or obligations in the morning. This underscores that the sleep duration during weekdays are too short. As short sleep duration is known to be related to a range of impairments in terms of academic functioning (Dewald et al., 2010), overweight (Danielsen et al., 2010), depressive symptoms and mental health problems in general (Cousins et al., 2011), there are reasons to be concerned about the sleep habits of the majority of adolescents included in the present study.

There was a high rate of insomnia in the present study, ranging from 13.6% to 23.6%, depending on diagnostic criteria. The paucity of studies assessing insomnia in adolescents and the different definitions used hinders comparisons across studies, but the prevalence is higher than the estimated rates of between 4 and 10% reported in previous studies using DSM-IV insomnia criteria (Johnson et al., 2006; Ohayon et al., 2000). Another potential limitation in the DSM-IV definitions concerns the independence of co-occurring psychiatric disorders. While the DSM-IV differentiates between primary insomnia and insomnia related to another disorder, we have not differentiated between these subtypes in order to ease comparison with the other diagnostic systems. Another limitation of the present study relates to the exclusive use of questionnaire-based information. Thus, diagnostic interview data, the gold standard for diagnosing insomnia, was not collected, but we relied instead upon a broad range of questionnaire-based sleep parameters that were used in accordance with specific diagnostic definitions. The criteria for daytime functional impairment in the present study were tiredness and sleepiness assessed by a joint variable. Although sleepiness is used more commonly as a symptom of obstructive sleep apnea, we chose to include both tiredness and sleepiness in the operationalization of insomnia due to a large overlap of these terms in the Norwegian language and due to limited ability in lay-people to discriminate between the two constructs.

A final limitation relates to how sleep duration was calculated. No data on time awake in bed prior to putting the light out and lying awake prior to rising were collected, and thus TIB – (SOL + WASO) might also include such periods. This may, potentially, lead to sleep duration being even shorter than reported in the current study.

The present study has shown that prevalence rates are influenced strongly by the quantitative operationalization and definitions of insomnia. While including a cutoff for SOL and a longer duration of the symptoms based on the quantitative criteria for insomnia, nevertheless a high percentage (13.6%) was diagnosed with insomnia. While this could mirror the high rate of true insomnia in adolescents, it could also reflect the need for adjustment of some of the criteria related to the respondents' age/maturation, for instance. The cutoff (>30 min) for SOL used in the research criteria is based on epidemiological studies of adults. In adolescence, a long SOL seems to be the norm and may reflect pubertal-related delayed circadian rhythms (Taylor et al., 2005) as well as pubertal slowing of the homeostatic sleep drive build-up (Jenni et al., 2005). Thus, the appropriate cutoff for SOL in determining insomnia in adolescents has still not been settled. Nocturnal awakenings seems to be less frequent in adolescents, thus the present criterion (>30 min) might be acceptable. As the present study indicates, the duration criteria will have only minor effects on the prevalence rates as most of the adolescents report insomnia for more than 1 year. Based on the present study, we cannot decide what is the most accurate or optimal operationalization. How these definitions are related to outcome measures in terms of functioning/impairment, other co-occurring disorders and the associated level of distress should, in future, be used as guidelines for choosing the most useful definitions.

Gender differences emerged regarding sleep patterns and insomnia prevalences irrespective of insomnia definition, showing a considerable female preponderance. This is in accord with findings in some previous studies, especially the gender-typical pattern of higher insomnia rates after puberty in girls (Johnson et al., 2006). While the results support the importance of gender-specific analysis in this age group, the mechanisms leading to these differences are, by and large, unknown. While the shorter TIB and shorter sleep duration found in boys could be due to the lower perceived sleep need that they reported in the present study, the differences could also be related to later pubertal development in boys, as there was an interaction effect of gender and age on some sleep patterns; e.g. bedtime and rise time during weekdays. However, there was no interaction effect between age and gender on insomnia and sleep efficiency, and thus pubertal developmental levels are less likely to account for these differences. As insomnia and lying awake in bed has been found to be related to worrying and depressive symptoms, some of the gender differences in sleep efficacy, SOL and WASO may be related to the parallel increased rate of depression in girls after puberty (Danielsson et al., 2012).

The strengths of the present study include the combination of large sample size and inclusion of a broad range of sleep parameters. The attrition from the study could affect generalizability, with a response rate of approximately 53% and with adolescents in schools over-represented. Based on previous research from the former waves of the Bergen Child Study, non-participants often have more psychological problems than participants (Stormark et al., 2008).

How can we help adolescents to achieve a sleep duration in agreement with their self-perceived need, and also their need according to recommended guidelines? The magnitude of the problem demonstrated in the present study indicates that short sleep duration is a public health issue. The results from previous school-based sleep education intervention studies for adolescents show positive effects on knowledge, but not on basic sleep habits (Cain et al., 2010; Moseley and Gradisar, 2009). In addition, the adolescents in the present study seem to be aware of the discrepancy between their obtained sleep during weekdays and their sleep need. In an Australian study, parent-set betimes were found to be related to longer sleep duration and improved daytime functioning, suggesting that parents may be key in improving adolescents sleep (Short et al., 2011). However, as this was an observational study, further intervention studies are needed. Few older adolescents will probably regard such an approach as acceptable, as fewer than 10% of high school students seem to have their bedtimes set by parents (Carskadon and Acebo, 2002). On a societal level, later school starting times have been suggested as a means of improving adolescents sleep (Carrell et al., 2011; Danner and Phillips, 2008; Kirby et al., 2011; Vedaa et al., 2012). More research is needed to both pinpoint the mechanisms leading to short sleep, including the effect of electronic media (Cain and Gradisar, 2011). The current findings emphasize that sleep problems among adolescents are a significant public health concern, and that low-threshold interventions and prevention programmes should be targeted for this age group.

Conflicts of interest

No conflicts of interest declared.

Ancillary