Dysregulated sleep–wake cycles in young people are associated with emerging stages of major mental disorders

Authors

  • Elizabeth M. Scott,

    1. Clinical Research Unit, Brain & Mind Research Institute, The University of Sydney, New South Wales, Australia
    Search for more papers by this author
    • The first two authors contributed equally to the manuscript.
  • Rébecca Robillard,

    1. Clinical Research Unit, Brain & Mind Research Institute, The University of Sydney, New South Wales, Australia
    Search for more papers by this author
    • The first two authors contributed equally to the manuscript.
  • Daniel F. Hermens,

    1. Clinical Research Unit, Brain & Mind Research Institute, The University of Sydney, New South Wales, Australia
    Search for more papers by this author
  • Sharon L. Naismith,

    1. Clinical Research Unit, Brain & Mind Research Institute, The University of Sydney, New South Wales, Australia
    Search for more papers by this author
  • Naomi L. Rogers,

    1. Concord Medical School, Concord Centre for Cardiometabolic Health in Psychosis, Concord, The University of Sydney, New South Wales, Australia
    Search for more papers by this author
  • Tony K. C. Ip,

    1. Concord Medical School, Concord Centre for Cardiometabolic Health in Psychosis, Concord, The University of Sydney, New South Wales, Australia
    Search for more papers by this author
  • Django White,

    1. Clinical Research Unit, Brain & Mind Research Institute, The University of Sydney, New South Wales, Australia
    Search for more papers by this author
  • Adam Guastella,

    1. Clinical Research Unit, Brain & Mind Research Institute, The University of Sydney, New South Wales, Australia
    Search for more papers by this author
  • Bradley Whitwell,

    1. Clinical Research Unit, Brain & Mind Research Institute, The University of Sydney, New South Wales, Australia
    Search for more papers by this author
  • Kristie Leigh Smith,

    1. Clinical Research Unit, Brain & Mind Research Institute, The University of Sydney, New South Wales, Australia
    Search for more papers by this author
  • Ian B. Hickie

    Corresponding author
    1. Clinical Research Unit, Brain & Mind Research Institute, The University of Sydney, New South Wales, Australia
    • Corresponding author: Professor Ian B. Hickie, Brain & Mind Research Institute, The University of Sydney, Level 4, 94 Mallett St, Camperdown, NSW 2050, Australia. Email: ian.hickie@sydney.edu.au

    Search for more papers by this author

Abstract

Aim

To determine if disturbed sleep–wake cycle patterns in young people with evolving mental disorder are associated with stages of illness.

Methods

The sleep–wake cycle was monitored using actigraphy across 4 to 22 days. Participants (21 healthy controls and 154 persons seeking help for mental health problems) were aged between 12 and 30 years. Those persons seeking mental health care were categorized as having mild symptoms (stage 1a), an ‘attenuated syndrome’ (stage 1b) or an ‘established mental disorder’ (stage 2+).

Results

The proportions of individuals with a delayed weekdays sleep schedule increased progressively across illness stages: 9.5% of controls, 11.1% of stage 1a, 25.6% of stage 1b, and 50.0% of stage 2+ (χ2 (3 d.f.) = 18.4, P < 0.001). A similar pattern was found for weekends (χ2 (3 d.f.) = 7.6, P = 0.048). Compared with controls, stage 1b participants had later sleep onset on weekends (P = 0.015), and participants at stages 1b and 2+ had later sleep offset on both weekdays and weekends (P < 0.020). Compared with controls, all participants with mental disorders had more wake after sleep onset (P < 0.029) and those at stages 1a and 2+ had lower sleep efficiency (P < 0.040). Older age, medicated status and later weekdays sleep offset were found to be the three strongest correlates of later versus earlier clinical stages.

Conclusions

In relation to clinical staging of common mental disorders in young people, the extent of delayed sleep phase is associated with more severe or persistent phases of illness.

Introduction

Increasing evidence indicates that sleep disturbances are intricately linked to the onset, comorbidity and chronicity of common mental disorders.[1-3] Significant sleep disturbances, such as shortened or extended sleep duration, increased sleep fragmentation and shifts in the timing of sleep periods (phase advance or phase delay) occur in most disorders, including common anxiety, psychotic and affective syndromes. In particular, altered patterns of sleep and daytime activity are clearly associated with the onset and persistence of psychological distress in young people.[4]

Given the likelihood that sleep and circadian disturbances are risk factors to a range of disorders, there is a need to utilize more objective markers of these disturbances and to assess their relationships with illness type, severity, course and response to treatment. Habitual sleep–wake patterns, as measured by ambulatory monitoring, can potentially provide relevant information. For example, we have noted that the normal developmental shift towards later sleep onset and offset times in teenagers and young adults is not only more pronounced in those with depressive disorders, but that phase delay is more marked in those with bipolar illness types.[5] In young people with emerging psychiatric disorders, various illness markers may be evident before full-threshold symptoms are apparent.

It has been increasingly recognized that the early phases of major mental disorders are characterized by rather non-specific symptoms including anxiety, depressed mood, sleep disturbances and brief psychotic phenomena.[6-10] To better account for the progression from sub-syndromal, attenuated or mixed psychiatric syndromes to more formal major disorders, we have proposed the potential utility of a clinical staging model.[11] In this context, staging is an adjunct to formal diagnostic systems.

In the model, stage 1a refers to help-seeking individuals with non-specific symptoms and mild psychosocial dysfunctions. At stage 1b (‘attenuated syndromes’), symptoms become specific but attenuated and psychosocial dysfunctions are at least moderate. Stages 2 (‘discrete disorders’) and later are marked by clear illness episodes with persistent moderate to severe symptoms and a major impact on psychosocial functioning. Although stage 1a and 1b are considered to reflect ‘attenuated’ syndromes, participants may already meet full-threshold criteria for a mood or anxiety disorders according to a Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) or International Classification of Diseases, 10th Edition (ICD-10) diagnosis. By contrast, those rated as stage 2 or above will not only have a formal diagnosis but also be characterized by illness severity, persistence or recurrence.

This clinical staging model is designed to assist with early screening, prognostic assessment and treatment planning with a strong focus on secondary prevention (i.e. treating one phase of a condition to prevent the development of another major condition). Within the model, it is assumed that those at earlier stages may be effectively managed with less toxic or short-term treatments, with a strong emphasis on the use of non-pharmacological interventions and reduction of other known risk factors (e.g. concurrent substance misuse, social isolation, avoidant behaviour, dysfunctional thoughts or attitudes). We have reported the characteristics of this model (cross-sectionally and longitudinally) when applied to young people (aged 12–30 years) presenting for mental health care.[7]

To test the validity of this clinical model, and particularly the proposed delineation of attenuated (Stages 1a and b) from full-threshold (Stage 2 and above) disorders, there is a clear need to determine whether more objective markers can distinguish early versus later illness phases. To date, we have evaluated a number of neurobiological strategies including structural magnetic resonance imaging,[12] neuropsychological functions[13] and evening melatonin secretion.[14] Each has provided preliminary support for the differentiation of stages 1 and 2. In this study, we set out to determine the extent to which sleep–wake cycle disturbances and sleep phase delay are characteristic of young people with more severe or established disorders, independent of formal diagnostic grouping.

Methods

Participants

One hundred seventy-five young persons aged between 12 and 30 years participated in this study (see Table 1 for demographic characteristics). Of these, 21 healthy controls who reported no history of mental illness were recruited from the community and 154 help-seeking patients with emerging mental disorders were recruited from early intervention services for mental health problems in young people (Youth Mental Health Clinic at the Brain & Mind Research Institute (BMRI); and headspace, Campbelltown, Sydney, Australia[15, 16]). For 126 of these participants, data about occupational status were obtained and used to classify each participant in one of three categories: (i) not studying nor working, (ii) studying or working part-time, (iii) studying full-time, working full-time or studying and working concurrently (Table 1). For 116 participants from the patient groups, data about medication were available and were used to define two subgroups: those taking psychotropic medications and those who did not. Sixty-nine percent of the patients were taking psychotropic medication for their mental health condition (Table 1). Potential participants were excluded had they done shift work or travelled across more than two time zones within 60 days previous to data collection. Other exclusion criteria were history of stroke; neurological disorder; head injury with loss of consciousness of at least 30 minutes; medical condition known to affect cognition; or other psychiatric illness. All participants gave written informed consent prior to participation in the study. The study protocol was approved by the University of Sydney Human Research Ethics Committee.

Table 1. Sample demographic characteristics
 ControlsStage 1aStage 1bStage 2+
  1. *Stage 1a and 1b participants were significantly younger compared to Controls and Stage 2+ participants (F = 16.4, P < 0.001, all contrasts P < 0.001).
  2. Ctrl, controls; Full-time: studying full-time, working full-time or studying and working concurrently; NA, not applicable; Part-time: studying or working PT; S1, stage 1; S2+, stage 2+; SD, standard deviation.
n21188254
Mean age (SD)*24.4 (3.1)17.6 (4.0)19.1 (4.1)22.4 (4.3)
Gender    
% females57.1%61.1%46.3%50.0%
Occupational status    
% Not working nor studying0.013.311.733.3
% Part-time0.06.718.319.0
% Full-time100.080.070.047.6
Medication    
% currently medicatedNA61.174.997.5
% antidepressantsNA34.355.966.8
% sedativesNA5.90.00.0
% adjunctiveNA23.515.615.1
% mood stabilizersNA23.56.622.6
% antipsychoticsNA23.214.762.2
% benzodiazepinesNA5.62.611.5
% stimulantsNA11.814.43.8
% anticonvulsantsNA0.00.05.8
Lenght of actigraphy monitoring    
Mean days (SD)8.8 (2.8)13.7 (1.4)12.5 (3.6)12.6 (2.7)

Procedures

Clinical assessment

As described elsewhere,[7, 16] patients entering these youth-focused mental health services were assessed and managed by trained health professionals. All participants were assessed by a senior clinician (psychiatrist or clinical psychologist) and by a neuropsychologist using DSM-IV criteria and the BMRI Structured Interview for Neurobiological Studies. Patients were also assessed using detailed criteria developed for formal application of our clinical staging framework.[7] Clinical stages were assigned by two psychiatrists with extensive clinical and research expertise in youth mental health and our staging system (ES, IH). Our previous findings indicated an acceptable interrater reliability of independent ratings (k = 0.71).[7] Patients were rated as having either mild symptoms (stage 1a), an ‘attenuated syndrome’ (stage 1b) or an ‘established disorder’ (stage 2 and above, herein referred to as stage 2+). Eighteen patients were classified as being in stage 1a, with non-specific symptoms of anxiety or depression. Of these participants, clinicians also assigned comorbid diagnoses of mood disorders (n = 9), anxiety disorders (n = 4), developmental brain disorders (n = 3), chronic fatigue syndrome (n = 1) and oppositional defiant disorder (n = 1). Eighty-two patients were classified as being in stage 1b, with more significant anxiety, depressive or psychotic symptoms. Of these participants, clinicians also assigned comorbid diagnoses of mood disorders (n = 54), anxiety disorders (n = 13), psychotic disorders (n = 4), developmental brain disorders (n = 8), substance abuse (n = 2) or impulse control disorder (n = 1). Fifty-four patients were classified as being in stage 2+ with significant anxiety, mood or psychotic symptoms. Of these participants, clinicians also assigned comorbid diagnoses of mood disorder (n = 33), anxiety disorder (n = 6) or psychotic disorder (n = 15). Stage 1a and 1b patients were significantly younger compared with controls and stage 2+ patients (F = 16.4, P < 0.001, all contrasts P < 0.001).

Sleep–wake ambulatory assessment

Sleep–wake measures were obtained from 4 to 22 days of sleep diary and actigraphy monitoring (Actiwatch-64/L/2, Philips Respironics, Bend, OR, USA). The number of days for which actigraphy data was collected was significantly lower in the control group compared with all other groups (F(3,171) = 12.5, P < 0.001; see Table 1 for means).

Actigraphy is a well-recognized objective tool for the ambulatory measurement of sleep–wake patterns. It uses a multidirectional accelerometer system in a wrist monitor to record movement and to infer the timing of sleep and wake states. In healthy adult, actigraphy shares a minute-by-minute agreement rate of about 90% with polysomnography (sleep electroencephalography).[17] Actigraphy does not measure vigilance states based on cortical activity and therefore provides indirect estimations of sleep measures. Nevertheless, to simplify the terminology in the current manuscript, sleep variables estimated from actigraphy will be defined as ‘sleep’ and ‘wake’. Data from Actiwatch-L was collected over 1-minute epochs, and data from all other actimeter models was collected over 30-second epochs. Only the main (i.e. longer) sleep episode from each 24-hour period was included in the analysis. Rest onset/offset were automatically defined with Actiware 5.0 software (Philips Respironics) and were subsequently inspected visually by trained technicians using sleep diaries. Rest intervals were subsequently submitted to dual integration to define periods spent ‘awake’ and alseep, using a wake threshold value of medium sensitivity (40.0 activity counts/epoch).

The following sleep variables calculated from actigraphy data were averaged for week nights (Sunday to Thursday) and weekend nights (Friday and Saturday): sleep onset/offset (moments when it was estimated that participants fell asleep and woke up; mean timing of the onset/offset of the sleep episode), time in bed (total duration of the sleep episode; period between sleep onset and sleep offset), total sleep time (amount of time scored as ‘sleep’ within the sleep episode), wake after sleep onset (WASO: amount of time scored as ‘wake’ within the sleep episode) and sleep efficiency (percentage of time spent asleep during the time in bed period). Sleep onset latency was not calculated (nor integrated in sleep efficiency) because of insufficient reliable bedtime data across the sample (i.e. some of the actiwatch models used have no event markers and several participants did not return their sleep diaries). Herein, sleep efficiency can thus be considered to be mainly reflective of sleep consolidation (i.e. a concept reflecting sleep continuity across the night, with better sleep consolidation translating into more time spent asleep during the sleep episode as opposed to sleep fragmented by extended or multiple night time awakenings).

Actigraphy data were also used to identify participants with a delayed sleep phase profile. Recent epidemiologic data in young people between the ages of 19 and 24 years indicated an average sleep onset time of 23:17 h (±1:05) and a mean offset time of 7:25 h (+1:22).[18] In the current study, a delay of approximately two standard deviations from these means was considered to be a significant deviation from normal sleep–wake patterns: participants with a mean sleep onset later than 1:30 h and/or a mean sleep offset later than 10:00 h were categorized as having a delayed sleep profile.

Statistical analyses

All statistical analyses were conducted with Statistica 6.1 software (StatSoft Inc., Tulsa, OK, USA). Chi-square tests were used to compare the proportions of those who presented a delayed sleep phase across these groups. Six two-way multivariate analysis of covariances controlling for age and number of actigraphy days were conducted to compare each actigraphy variable on weekday and weekend days across the four clinical groups. Fisher's least significant difference tests were conducted for multiple comparisons of means on significant main effects. Significant interactions were decomposed using contrast analyses. Logistic regressions were done with clinical group as the dependant variable, and sleep variables (week and weekend), age, gender, medication status and occupational status as independent predictor variables. A pairwise approach was used to account for missing data.

Results

Proportions of individuals with delayed sleep phase

For week nights, there were significant group differences in the proportions of individuals who had a delayed sleep phase schedule, with later stage of illness being associated with higher rates of delayed sleep phase (Fig. 1). The frequency of delayed sleep schedules on weekdays were similar in control participants (9.5%) and stage 1a participants (11.1%), but increased progressively at stage 1b (25.6%) and stage 2+ (50.0%) (χ2 (3 d.f.) = 18.4, P < 0.001). A similar pattern was found for the weekends, with similar proportions of individuals with delayed profiles in control (28.6%) and stage 1a participants (22.2%), and higher proportions in stage 1b (48.8%) and stage 2+ (51.9%) participants (χ2 (3 d.f.) = 7.6, P = 0.048).

Figure 1.

Percentages of individuals with a delayed sleep phase profile in each stage group. Results from the chi-square analyses indicated significant differences for week (χ2 (3 d.f.) = 18.4, P < 0.001) and weekend days (χ2 (3 d.f.) = 7.6, P = 0.048). (image) week, (image) weekend.

Group differences

Table 2 and Figure 2 summarize group differences in continuous sleep variables for week nights and weekend nights after controlling for age and the number of days of actigraphy monitoring. Significant group by time interactions were found for sleep onset and sleep offset. Compared with control participants, stage 1b participants had significantly later sleep onset on weekend nights (P = 0.015), but not on week nights (P = 0.187). Compared with control participants, stage 1b and stage 2+ participants had significantly later sleep offset both on week and weekend nights, but these effects were more pronounced on week nights (both P < 0.001) than on weekend nights (P = 0.002 and P = 0.020, respectively). Significant group effects were found for WASO and sleep efficiency. WASO was higher in all patient groups than in controls (all contrast P < 0.029). Sleep efficiency was significantly lower in stage 1a and 2+ participants compared with controls (P = 0.002 and P = 0.040, respectively) and tended to be lower in stage 1b participants than in controls (P = 0.066). There was no other significant group difference.

Table 2. Sleep–wake patterns across stage groups
 Mean (SD)GroupWeek periodGroup*time
ControlsStage 1aStage 1bStage 2+
WeekWEWeekWEWeekWEWeekWEFPPartial eta2FPPartial eta2FPPartial eta2
  1. Means, standard deviations (SD) and statistics from MANCOVAs controlling for age and the number of days of actigraphy.
  2. For all significant time effects, weekend values were higher than week values (see text for detailed time and group*time effects).
  3. Ctrl; controls (n = 21); S1a, stage 1a (n = 18); S1b, stage 1b (n = 85 for weekend data and 83 for week data); S2+, stage 2+ (n = 55); SE, sleep efficiency; SleepON/OFF, sleep onset/offset; TiB, time in Bed; TST, total sleep time; WASO, wake after sleep onset.
SleepON23:25 (1:09)0:03 (2:05)23:32 (1:49)0:16 (2:00)23:55 (1:35)0:50 (1:57)0:11 (1:39)0:24 (1:38)3.70.0140.06
SleepOFF7:33 (1:23)8:49 (1:30)8:05 (1:37)9:08 (1:36)8:51 (1:38)9:59 (1:49)9:16 (1:31)9:43 (1:40)3.10.0290.05
TiB488.2 (69.6)526.5 (106.8)513.3 (49.1)531.5 (90.9)535.6 (74.7)549.8 (85.7)544.9 (87.4)558.2 (114.5)2.00.1240.033.0.0860.021.20.3140.02
TST433.0 (60.4)461.2 (86.1)431.9 (47.7)427.1 (67.5)458.3 (63.6)472.4 (75.6)468.7 (87.5)475.1 (109.3)2.20.0950.041.6.0950.011.30.2760.02
WASO55.2 (25.0)65.3 (37.6)81.4 (31.7)84.7 (38.0)76.0 (31.6)77.4 (29.2)76.3 (31.9)81.1 (36.8)3.90.0100.061.5.2280.012.30.0850.04
SE88.9 (4.1)87.8 (5.0)84.2 (5.6)81.3 (10.7)85.8 (5.5)86.0 (4.9)85.8 (5.9)84.9 (7.2)5.20.0020.091.8.1820.012.60.0510.05
Figure 2.

Mean and standard error of the mean of sleep parameters showing significant effects; results from the MANCOVAs. Top panels: group by week period interactions for sleep schedule; lower panels: group effects for sleep consolidation variables. *P < 0.050, **P < 0.001. (image) week, (image) weekend.

Predictive factors for clinical stage

A model integrating age, medication status, occupational status, week sleep offset and time in bed, and weekends sleep onset and sleep efficiency resulted from multiple regression analyses (full model: R = 0.53, F(7,76) = 4.3, P < 0.001). Age and medication status were the two strongest independent predictors for clinical stage, uniquely accounting for 4.8% (β = 0.25, P = 0.026) and 4.9% (β = 0.26, P = 0.025) of the total variance in clinical stages (with later clinical stages being associated with older age and medicated status). Later week sleep offset was the only other significant independent predictor of later clinical stages, explaining 4.1% of the total variance (β = 0.32, P = 0.040).

Discussion

The current study indicates that the timing of the sleep–wake cycle changes in association with the development of early stages of common mental disorders. Delayed sleep schedule was associated with ‘attenuated’ and more ‘discrete’ stages of illness in young persons with emerging disorders. The effects were generally more notable on week nights than on weekend nights. Compared with healthy controls, all mental disorder subgroups had significantly higher WASO, whereas stage 1a and 2+ participants had lower sleep efficiency, suggesting that most participants with emerging disorders have difficulties in maintaining consolidated sleep across the night.

Interestingly, the difference in sleep efficiency was more pronounced at stage 1a (i.e. the earliest phase; help seeking with non-specific illness symptoms and only mild psychosocial dysfunctions). It has been proposed that poor sleep consolidation can arise from dysfunctions of wakefulness inhibition mechanisms linked to a fight-or-flight systemic response to an acute stressor.[19] It is possible that the emergence of new anxiety or depressive symptoms perturb these wake inhibition mechanisms, leading to sleep fragmentation very early in the course of common mental disorders. As the disorder then becomes more persistent or severe, oversleeping may emerge as a compensatory mechanism for accumulated sleep debt. It is also possible that primary dysfunction of the sleep–wake cycle may be driving the emergence of both sleep fragmentation and subsyndromal psychiatric symptoms.

The rates of sleep phase delay were more frequent in stage 1b and 2+ participants, with rates being highest in those with more persistent or severe disorders. Correspondingly, sleep offset was significantly delayed in both the stage 1b and 2+ groups compared with the control group. This is in line with our previous finding that young people with later stages of illness have lower levels of melatonin in the evening, a physiological state likely to be less conducive to early sleep onset.[14] Furthermore, we previously observed that compared with those with unipolar depression, young people with bipolar disorders (commonly seen as a more complex clinical entity) had further delayed sleep schedules and melatonin secretion profiles.[5, 20] To better understand the relationship between sleep–wake disturbances and mental illness stages, future longitudinal studies should investigate how various illness trajectories relate to changes in the sleep–wake cycle.

Importantly, several lines of evidence suggest that delayed sleep patterns or later chronotypes, especially in young persons, are associated with increased depressive symptoms and higher suicidality.[21-24] As we observed previously,[7] patients at stage 1a and 1b were younger than those at stage 2+, but all the differences reported remained significant after controlling for age. Occupational and social participation often changes in the course of emerging mental disorders. Such social parameters may, in turn, have a strong influence on the timing of the sleep–wake cycle. Occupational status is thus one of the possible mediating factors between the development of abnormal sleep schedules and the emergence of more severe or persistent mental disorders. In cross-sectional studies such as the one described here, it is not possible to determine the direction of causality between changes in sleep–wake cycle timing and the emergence of later phases of mental disorders. Although it is possible that more delayed sleep offset (and delayed circadian phase) in later clinical stages are a consequence of loss of social constraints over sleep timing (i.e. work or school imposed early morning starting times), it is also likely that more dysregulated sleep–wake cycles drive greater social and occupational impairment. Importantly, sleep offset time on week nights was found to be a significant predictor of illness stage independently of occupational status.

Other factors such as medication usage may also be contributing to the association between altered sleep–wake cycle patterns and later phases of illness (given the higher rate of medication use in those at later stages of illness). Although sleep onset time, WASO and sleep efficiency varied across clinical groups, they were not significant independent predictors of clinical stage in a model controlling for age, occupational status and medication status. However, later sleep offset times on week nights (when occupational status is likely to be more influential) were found to be predictive of later stages of illness independently of age, medication status and occupational status.

Importantly, independent of the direction of causation with other social or illness-dependent factors, the sleep disturbances found in these patients are likely to have very negative impacts on their well-being. Additionally, they represent potential intervention targets to improve global clinical profiles, including both mental and physical health outcomes. It has long been proposed that phase shifting the sleep–wake cycle to earlier times can help in alleviating common depressive symptoms.[25, 26] Interventional studies are required to determine whether strategies aiming to restore the timing of the sleep–wake cycle may be a relevant intervention pathway in young persons with psychiatric disorders.[27-29] Notably, if changes in occupational status are one of the main factors modulating the relationship between dysregulated sleep–wake cycles and illness severity, interventions aiming to restore regular daytime schedules are a high priority for further systematic evaluation.

The current results need to be interpreted according to some methodological limitations. Although we attempted to control for several factors potentially influencing the nature and intensity of sleep disturbances across stages of illness, information about medication and occupational status was not available for all participants. Although we controlled for it in some of the analyses, the length of actigraphy monitoring period varied across the sample and included a limited number of weekend days, which could have influenced the lower sensitivity of weekend days. As the data presented are cross-sectional, we cannot determine the longitudinal development of these patterns of change in the sleep–wake cycle or their capacity to predict illness course or treatment response. As actigraphy does not directly measure the brain's transitions between the wake and sleep states, it only provides indirect evidence of the actual sleep–wake patterns.

Based on objective data collected in a large cohort of young persons with emerging mental disorders, this study suggests that (i) the later stages of ‘attenuated’ syndrome and discrete disorders are associated with more significant sleep phase delays; (ii) a range of sleep–wake cycle changes are influenced by age, medication and occupational status; and (iii) delayed sleep offset on week nights may emerge as an independent predictor of later clinical stages. Future longitudinal and intervention studies are required to investigate the causal relationships between delayed sleep schedules, illness progression and global functioning.

Acknowledgements

This study was funded from a National Health and Medical Research Council (NHMRC) Australia Fellowship awarded to IBH. Associate SLN is funded by an NHMRC Career Development Award. DFH is currently supported by a grant from the NSW Health, Mental Health and Drug & Alcohol Office. RR received a postdoctoral training award from the Fonds de la recherche en santé du Québec.

Ancillary