Developmental trajectories of sleep during childhood and adolescence are related to health in young adulthood

Sleep behaviour is correlated and causally related to physical and mental health. Limited longitudinal data exist on the associations of poor sleep behaviour in childhood and adolescence with adult health. Parent‐reported sleep behaviours from 1993 participants of the Raine Study (at ages 5, 8, 10, 14, 17) were used to determine sleep trajectories (using latent class growth analysis).


| BACKG ROU N D
Poor sleep has been associated with accidental injuries and death, poorer academic performance, more negative moods and poorer physical health in young people 1 and multiple physical and mental outcomes in adulthood, such as depressive mood and poorer physical health. 2  Some evidence suggests poor sleep behaviours during early childhood may have multiple consequences in later life, including attentional difficulties, poor impulse and emotional control, metabolic dysfunction and substance-related problems in young adulthood. 7,8 However, current evidence, which is mostly based on sleep problems being measured at very few time points, does not capture the developmental changes in sleep that occur during childhood and adolescence and their relationship to later health outcomes. 9 One study examined problematic sleep, over childhood and adolescence (ages 2-15 years) and, using structural equation modelling, showed that poor sleep behaviours in the younger years were associated with risk-taking behaviour during adolescence. 10 More recently, trajectories of sleep problems from 5 to 14 years of age have been shown to relate to emotional problems over the same period, 5,6 however, health outcomes in adulthood were not examined.
These studies suggest that the long-term patterns of sleep problems over childhood may have bearing on health outcomes in adulthood. Therefore, this study aimed to investigate whether trajectories of sleep behaviours over the critical developmental periods of childhood and adolescence were associated with physical and mental health outcomes in young adulthood.

| Participants
Participants were from the Raine Study (www.raine study.org. au). The study has been described in detail elsewhere. 11

| Sleep Trajectories
Parents reported on their children's sleep problems at ages 5, 8, 10 and 14 and 17 using the Child Behaviour Checklist (CBCL, 13 ). The CBCL was administered at each of the 5 time points, from which six items were used to create a composite score of sleep problems at each time point. The items included 'trouble sleeping', 'nightmares', 'overtired without good reason', 'sleeps less than most kids', 'talks or walks in sleep' and 'sleeps more than most kids during day and/or night'. Each item was rated on a 3-point scale (0 = not true, 1 = somewhat or sometimes true, 2 = very true or often true) and a sum score of the 6-item (range 0-12) was used to represent the level of poor 2 (SF-12) was used to assess self-rated health and well-being at age 20. 16 The SF-12 consists of 12 Likert questions that were reversecoded and transformed into a 100-point scale using standardised guidelines. 17 From these items, individual scores from four items were used to create the Physical Component subscale, with higher scores indicating better health. 17  Health Survey version 2 (SF-12) was used to assess self-rated mental health and well-being 18 with higher scores indicating better health.
Mental health was also assessed with the 21-item self-reported Depression Anxiety Stress Scales (DASS-21). 19 The DASS-21 has been validated in clinical and non-clinical populations. 20 Participants were asked to rate the extent to which they had experienced each state over the past week on a four-point severity/frequency scale with responses ranging from 0 (did not apply to me at all) to 3 (applied to me very much or most of the time). The DASS-21 yields separate depression, anxiety and stress subscale scores (scores range from 0 to 21, based on 7-item), and a composite total score that is the sum of the three subscales (range 0-63, based on all 21-items). Although cut-off scores defining mild/moderate/severe and extremely severe have been developed, the DASS is based on a dimensional concept of psychological disorders, as opposed to a categorical concept. 19 Therefore for this study, continuous scores were used in the analysis.

| Other measures at age 5 and age 20
Selected additional measures from early childhood and young adulthood (known to be associated with poor physical and mental health outcomes i.e., BMI, family income, education, smoking, sleep disturbances, sedentary time and physical activity) were reported and compared between the trajectories (see Table 1).
Variables that were significantly different between the trajectories were then used as covariates in subsequent models. At the 5-year follow-up, the participant's height (m) and mass (kg) were measured, and a BMI Z-score was calculated and converted into a percentile (relative to the participant's sex and age). Also, at

| Data analysis
To determine the trajectories of sleep problems, latent class growth analysis (LCGA) 23 (Table S1), with model fit statistics (for sex-specific (Table S2) and other combinations of time points) presented in supplementary materials.
Sensitivity analyses were explored including for participants with one missing time point (Table S3) and for those who had no missing time points (Table S4).
Generalised linear or negative binomial regression models were used to determine the associations between the sleep trajectory classes and physical and mental health outcomes in adulthood.
Models were weighted according to the probability of membership of the trajectory class (predictor variable) and adjusted for sex. To consider the early and contemporaneous effect of selected variables associated with physical and mental health, we ran additional models adjusting for the following variables: family income at age 5, smoking and sleep disturbances at age 20. Interactions were explored between sex and trajectory membership with health outcomes and results are presented separately for males and females where the interaction was significant at p < 0.10. All analyses were conducted using Stata v13.1.

| Sleep problems
The frequency of parent reported CBCL sleep problems classified as 'very true/often true' are shown in

| Sleep problem trajectories
Model fit statistics (Table S1)

| Health outcomes between the trajectories
Because there was no significant sex by trajectory membership interaction, the association between trajectory membership and health outcomes was reported for females and males combined. Patterns of differences across the three trajectories remained similar when adjustments for sex, family income at age 5 and smoking and sleep disturbance (at age 20) were made ( Table 3). The differences in physical and mental health outcomes across the three trajectories were similar in magnitude for both sex-only and fully adjusted models for most variables and are described as follows.

| Physical health outcomes
In the sex-only adjusted model, but not in the fully adjusted model,   Note: Generalised linear models all weighted for probability of membership and adjusted for sex (white rows) or adjusted for sex, family income at age 5; and sleep disturbance and smoking at age 20 (shaded grey rows). Bold font indicates significantly different differences at p < 0.05.
the childhood and adolescent years in the current study supports similar work which has demonstrated a continuation of poor sleep behaviours from adolescence to adulthood. 26 In the current study, better sleep behaviours across childhood and adolescence were associated with better young adult physical (better SF-12 scores and more favourable body composition) and mental health (better SF-12 scores and lower depression scores).
This finding builds on previous work which has found trajectories of reduced sleep duration across childhood to be related to poorer health outcomes in childhood. 8,27 For example, Magee et al. 27

CO N FLI C T S O F I NTE R E S T
None.

CO N S E NT TO PA RTI CI PATE
Parents initially provided informed, written consent to participate in the study until their children were old enough to provide their consent.

CO N S E NT FO R PU B LI C ATI O N
Parents initially provided informed, written consent to participate in the study until their children were old enough to provide their consent.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data access is subject to restrictions imposed in order to protect participant privacy. All researchers using Raine Study data must sign a data access agreement stipulating that data may not be released to anyone other than the investigators of the approved project.
Additional details regarding data access are available from: http:// www.raine study.org.au/.