Accepted Manuscript Title: Sleep deprivation leads to mood deficits in healthy adolescents Author: Michelle A. Short, Mia Louca PII: DOI: Reference:
S1389-9457(15)00698-X http://dx.doi.org/doi:10.1016/j.sleep.2015.03.007 SLEEP 2728
To appear in:
Sleep Medicine
Received date: Revised date: Accepted date:
3-11-2014 26-3-2015 27-3-2015
Please cite this article as: Michelle A. Short, Mia Louca, Sleep deprivation leads to mood deficits in healthy adolescents, Sleep Medicine (2015), http://dx.doi.org/doi:10.1016/j.sleep.2015.03.007. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Sleep deprivation leads to mood deficits in healthy adolescents
Michelle A. Short, PhD & Mia Louca, BPsycSci(Hons) Centre for Sleep Research University of South Australia
This study was performed at the Centre for Sleep Research, University of South Australia and funded by a Divisional Research Performance Grant from the University of South Australia. The authors have no conflicts of interest to disclose.
Corresponding author: Michelle A. Short Centre for Sleep Research GPO Box 2471 Adelaide SA 5001 PH: +61 8 8302 1966 Email: [email protected] 1 Page 1 of 23
Highlights
The effect of sleep deprivation on mood was examined in healthy adolescents Sleep loss causally affected all discrete mood states Males and females showed differential vulnerability to sleep loss Discrete mood states differed in their sensitivity to sleep loss Abstract
Study Objectives: To investigate the effects of 36 hours of sleep deprivation on the discrete mood states of anger, depression, anxiety, confusion, fatigue and vigour in healthy adolescents. Method: Twelve, healthy adolescent good sleepers (6 male), aged 14 to 18 years (M=16.17, SD=0.83), spent three consecutive nights in the sleep laboratory of the Centre for Sleep Research: two baseline nights with 10h sleep opportunities and one night of total sleep deprivation. Every two hours during wake they completed the Profile of Mood States – Short Form. Mood across two baseline days was compared to mood at the same clock time (0900h to 1900h) following one night without sleep. Results: The subscales of depression, anger, confusion, anxiety, vigour and fatigue were compared across days. All mood states significantly worsened following one night without sleep. Females showed a greater vulnerability to mood deficits following sleep loss, with greater depressed mood and anxiety following sleep deprivation only witnessed among female participants. While males and females both reported more confusion following sleep deprivation, the magnitude of this effect was greater for females. Conclusions: This study provides empirical support for the notion that sleep loss can causally affect mood states in healthy adolescents, with females having heightened vulnerability. Understanding the detrimental effects of insufficient sleep during adolescence is important as it is a stage where sleep loss and mood dysregulation are highly prevalent. These findings escalate the importance of promoting sleep for the wellbeing of adolescents at this critical life phase.
KEYWORDS: Adolescence, Anxiety, Depressed Mood, Anger, Fatigue, Sleep, Affect, Emotion.
2 Page 2 of 23
INTRODUCTION Sleep deprivation has been linked to serious changes in mood states and their regulation (Haack & Mullington, 2005). While experimental studies have begun to elucidate the relationship between sleep and mood in adults, this relationship is underresearched in adolescents. Indeed, the lack of experimental research examining the impact of sleep on mood in adolescents has been identified as a critical gap in the literature (Talbot, McGlinchey, Kaplan, Dahl, & Harvey, 2010). To address this gap, present study examines the effect of 36 hours of sleep deprivation on the discrete mood states of depression, anxiety, anger, fatigue, vigour and confusion in an adolescent sample. Evidence for the relationship between sleep disorders and mood disorders is strong in adults and adolescents (Baglioni et al., 2011; Harvey, 2008; Lovato & Gradisar, 2014; Taylor, Lichstein, Durrence, Reidel, & Bush, 2005). Indeed, sleep problems are one of the diagnostic criteria for major depressive disorder. However, it is unclear to what degree sleep loss can perturb mood in healthy adolescents. Establishing a causal link between sleep and mood in healthy populations would provide important support for the causal relationship from insufficient sleep to mood deficits, as well as highlighting the importance of preventative public health strategies targeting sleep and well-being in this age group. As mood disorders are one of the leading causes of morbidity and mortality in adolescents (Forbes & Dahl, 2005), this is an important and timely area of research. Previous work examining sleep deprivation in adults has elucidated a range of functional deficits associated with sleep, as well as potential biological mechanisms underpinning them. Adult research has shown mood deficits in response to both sleep deprivation and sleep restriction, including decreased vigour and increased confusion and fatigue (Dinges et al., 1997; C. L. Drake et al., 2001). Some studies have also found
3 Page 3 of 23
increased anger (Haack & Mullington, 2005), anxiety (Brendel et al., 1990; Reynolds, Kupfer, Hoch, & Stack, 1986) and depressed mood (Scott, McNaughton, & Polman, 2006; Selvi, Gulec, Agargun, & Besiroglu, 2007) in response to sleep loss, however these mood states have not been shown to consistently worsen (Dinges et al., 1997; C. L. Drake et al., 2001). While vigour, confusion and fatigue worsen with moderate levels of sleep restriction, mood states, such as depressed mood, anger and anxiety, tend to worsen with more severe sleep restriction, and in particular, following total sleep deprivation (Babson, Trainor, Feldner, & Blumenthal, 2010). This suggests a differential sensitivity to sleep loss between mood states. In addition, there may be a dose-dependent relationship between sleep loss and mood deficits, with greater sleep loss predicting greater mood deficits. This is consistent with the dose-response relationships between sleep duration and other functional outcomes (Van Dongen, Maislin, Mullington, & Dinges, 2003). As such, the present study utilises a condition of total sleep deprivation, rather than sleep restriction, to operationalise sleep loss in order to best establish a causal relationship between sleep loss and a range of discrete mood states. In addition, the sensitivity of different mood states to the same degree of sleep loss will be examined. Despite many excellent studies in adults, there are several reasons why these findings cannot be generalised to adolescents. Adolescents differ from adults in terms of brain maturation, sleep homeostatic and circadian system function, and psychosocial milieu (Crowley, Acebo, & Carskadon, 2007; Dahl, 2004; Jenni, Achermann, & Carskadon, 2005). Adolescents are argued to have fewer resources to cope with the affective challenges they face because of their developmental stage (Talbot et al., 2010). Sleep has shown to have a role in regulating emotional brain-states, with the amygdala implicated in the processing of emotionally salient information (Yoo, Gujar, Hu, Jolesz, & Walker, 2007). In an fMRI study, Yoo and colleagues (Yoo et al., 2007) found an 4 Page 4 of 23
amplified hyper-limbic response by the amygdala to aversive emotional stimuli under conditions of sleep deprivationand reduced functional connectivity between the amygdala and the medial prefrontal cortex (mPFC). The mPFC has an inhibitory function over the amygdala, meaning that heightened emotional reactivity related to amygdala function is not attenuated to the same degree by the mPFC during conditions of sleep loss. As the prefrontal cortex (PFC) is still developing in adolescents (Talbot et al., 2010), mood deficits arising from sleep loss may be amplified. While experimental research into the relationship between sleep and mood in adolescents is limited, several excellent studies have begun to explore this relationship in healthy populations (Baum et al., 2014; Dagys et al., 2012; Dewald-Kaufmann, Oort, & Meijer, 2014; McGlinchey et al., 2011; Talbot et al., 2010). Talbot and colleagues (Talbot et al., 2010) restricted sleep to 6.5 hours for the first night and 2 hours for the second night and then compared mood (using the Positive Affect and Negative Affect Schedule for Children, or PANAS-C) to a rested condition in which participants slept between 7 and 8 hours per night for two nights. This effect of sleep loss was investigated across early adolescents, mid adolescents and adults. It was found that while adolescents reported less positive affect when sleep restricted compared to when rested, negative affect did not change. Adolescents also reported higher levels of anxiety following a catastrophizing ask and rated the likelihood of the catastrophes much higher when sleep deprived (Talbot et al., 2010). A study by Baum and colleagues (Baum et al., 2014) found changes to both positive and negative mood states following sleep loss. The authors conducted a homebased experimental sleep restriction protocol in a sample of 50 healthy adolescents aged 14 to 17 years. Participants completed one week of baseline sleep, followed by 5 nights of 6.5 hours’ time in bed (TIB) night and 5 nights of 10 hours’ TIB in a counterbalanced order, 5 Page 5 of 23
with one weekend as a washout period in between conditions. Participants reported significantly more anxiety, anger, confusion and fatigue, and significantly less vigour on the Profile of Mood States during the sleep restriction period compared to the extended sleep week. Contrary to expectations, depressed mood did not change between conditions. While these adolescent sleep restriction studies cast some doubt over the ability of sleep loss to perturb depressed mood, another study using a sleep extension paradigm found that more sleep resulted in less depressed mood (Dewald-Kaufmann et al., 2014). Fifty-five adolescents were identified as having chronic sleep reduction and randomised them to either a sleep extension and sleep hygiene group or a control group. Time in bed was extended via earlier bedtimes over a two week period by one hour and monitored actigraphically. At the end of the two week period, those in the sleep extension group obtained more sleep and reported decreased depressed mood on the Children’s Depression Inventory . Unfortunately, other mood states were not assessed. While these studies are invaluable in elucidating possible causal relationships between sleep duration and mood, results have been inconsistent between studies and there are a number of methodological challenges yet to be addressed. For example, while Talbot and colleagues (Talbot et al., 2010) bought participants into the laboratory for the second night of sleep restriction, the remainder of the sleep restriction/extension studies in adolescents utilised at-home protocols. These have the advantage of providing greater ecological validity, but, as with any home-based sleep experiment, this comes at a possible loss to experimental control. Beebe et al., (2008) acknowledged a potential for nonadherence to study protocols within home-based sleep restriction studies due to the lack of behavioural monitoring of participants. In one adolescent sleep restriction study, 15% of adolescent participants reported non-compliance with the protocol (Jiang et al., 2011) and a number of participants dishonoured prohibitions against drinking caffeine or napping. 6 Page 6 of 23
Relying on self-reported adherence to protocols means that participants may under-report their napping or caffeine use (Beebe et al., 2008). Some studies also allowed limited consumption of caffeine (Baum et al., 2014; Dewald-Kaufmann et al., 2014). As caffeine can eliminate the mood and alertness deficits caused by sleep loss (Penetar et al., 1993), this may have minimised or even ameliorated sleep-related mood deficits. Finally, there is uncertainty over the degree to which confounding factors known to affect both mood and sleep, such as physical activity (Rothon et al., 2010) or exposure to bright light (Golden et al., 2005), are present. To address these methodological concerns the present study utilised a laboratory-based protocol. This study will add to our understanding of the effect of sleep loss on adolescent mood: (i) by establishing whether sleep loss can perturb a range of mood states in adolescents, including depressed mood, (ii) by establishing whether mood states differ in their sensitivity to sleep loss, and (iii) by establishing whether male and female adolescents differ in their mood response to sleep loss. Participant mood following one night of total sleep deprivation was compared to that across two baseline days following 10 hour sleep opportunities. It was hypothesised that there would be significantly increased depressed mood, anxiety, anger, fatigue and confusion and decreased energy following one night without sleep, when compared to two baseline periods when 10 hour sleep opportunities are given. Based on findings from adults sleep deprivation studies (Brendel et al., 1990; Dinges et al., 1997; C. L. Drake et al., 2001; Haack & Mullington, 2005; Reynolds et al., 1986; Scott et al., 2006; Selvi et al., 2007), it is hypothesised that mood changes in response to sleep loss will be significantly greater for the discrete mood states of confusion, vigour and fatigue than for anger, anxiety and depressed mood. Finally, there is some argument that mood deficits in response to sleep loss may be greater in females than males (Armitage & Hoffmann, 2001). However, 7 Page 7 of 23
due to the lack of previous research addressing sex differences across the range of mood states, these analyses were exploratory. METHOD Participants Participants included 12 adolescents (50% male) aged 14 to 18 years (M=16.17, SD=0.83) who attended secondary schools in South Australia. Participants were physically and psychologically healthy, as determined by parent- and self-reported surveys.. Participants who obtained less than 8 hours sleep per night, had average sleep onset latencies of more than 30 minutes per night, weekend bedtime delay of two hours or more or who were extreme morning or evening chronotypes (total scores ≤ 22 or ≥44) were excluded from the study, as were participants who scored of 16 or greater on the CES-D or received scores above “mild” on any of the subscales of the DASS-21. Participants did not use medication and were free of any medical, psychological and sleep disorders, as determined by parent and adolescent self-report questionnaires, a 7-day sleep diary and wrist actigraphy. Participants were either late- or post-pubertal (Tanner stages 4 or 5), as measured by the Pubertal Development Scale (Petersen, Crockett, Richards, & Boxer, 1988) and had normal or corrected-to-normal vision. Ethics approval was granted by the University of South Australia Human Research Ethics Committee and participants received an honorarium for their participation. Written, informed consent was obtained from each adolescent and a parent. Mood Measure The Profiles of Mood States Questionnaire-Short Form (POMS-SF) (Shacham, 1983) was used to measure mood. The POMS-SF consists of a list of 37 adjectives 8 Page 8 of 23
describing mood states that fall into one of six subscales: Anxiety (N = 6 items), Vigour (N = 6 items), Depressed mood (N = 8 items), Fatigue (N = 5 items), Anger (N = 7 items) and Confusion (N = 5 items) (Curran, 1995). Respondents rate the extent to which they are experiencing that mood state at that point in time on a 5-point Likert scale which ranges from 1 (not at all) to 5 (extremely). For each subscale, higher scores indicate greater presence of that mood. To enable comparison of relative impact of sleep loss on mood across the 36 hour period of continual wakefulness, and to account for differences in the number of items within each mood subscale, average scores were calculated for each subscale at each time point, including those across the night. Procedure The present study utilised a repeated measures design. The independent variables were day (baseline 1, baseline 2, sleep deprivation) and time (09:00h, 11:00h, 13:00h, 15:00h, 17:00h, 19:00h) and the dependent variables were Anxiety, Vigour, Depressed Mood, Fatigue, Anger and Confusion. Parents of participants underwent a telephone screening interview using The Sleep, Medical, Education and Family History Survey. Suitable participants were invited to attend an interview at the Centre for Sleep Research where the adolescents completed a screening battery Sleep, Medical, Education and Family History survey and a questionnaire battery containing the School Sleep Habits Survey (Wolfson et al., 2003), Centre for Epidemiological Studies Depression Scale (CES-D) (Radloff, 1977), Composite Morningness/Eveningness Scale (Smith, Reilly, & Midkiff, 1989), Depression Anxiety and Stress Scale (21 item; DASS-21) (Lovibond & Lovibond, 1995), and Pediatric Daytime Sleepiness Scale (C. Drake et al., 2003) and were given a one-week sleep diary 9 Page 9 of 23
and activity monitor to use over one typical school week. Three potential participants were excluded following screening due to long sleep onset latencies, heightened sleep variability or elevated anxiety. During the week prior to the study, participants maintained a regular sleep pattern, with bedtime at 2200h and wake time at 0700h (Roberts, Roberts, & Duong, 2009) in order to remove any sleep debt and ensure similar circadian entrainment between participants. Adherence was verified using sleep diaries, wrist actigraphy, and timestamped messages to the sleep laboratory morning and night. All participants who began the pre-study sleep schedule maintained adherence to the schedule and completed the study. Participants spent 3 consecutive nights in the Sleep Laboratory of the Centre for Sleep Research, from 1600h on day 1, until 2000h on day 4 (see Figure 1). The sleep laboratory is sound attenuated and light and temperature controlled. Light exposure was kept to less than 50 lux during all wake periods and less than 1 lux during sleep opportunities. Ambient temperature was 21°C (±1°C) at all times. Participants completed a neurobehavioural test battery (NTB) every two hours during wake. The NTB included a 10 minute psychomotor vigilance task, digit symbol substitution task, POMS-SF and the Karolinska Sleepiness Scale every two hours during wake period. NTBs began at 0900h to avoid the effects of sleep inertia on mood in the period immediately following wake. Results of the neurobehavioural measures are presented elsewhere (Louca & Short, 2014). On the first 2 baseline nights the 10-hour sleep opportunities were recorded using polysomnography (PSG), with a montage including frontal (F3/F4), central (C3,C4), and occipital (O1,O2) referenced against A1/A2, bilateral electrooculography, electromyography and electrocardiography. In between test bouts, participants could interact with fellow participants and research staff, watch DVDs, play card and board
10 Page 10 of 23
games and complete jigsaw puzzles. Continuous behavioural observation ensured that participants remained awake during all designated wake periods.
RESULTS
On average, participants obtained approximately 9 hours sleep during the two baseline sleep opportunities in the laboratory (X = 537.08 min, SD = 21.47), establishing the adequacy of the baseline nights to ensure participants were well-rested coming into the sleep deprivation night. Average sleep onset latencies of 25.60 minutes (SD = 15.58) per night indicated that participants were able to initiate sleep in the laboratory setting and were unlikely to be experiencing circadian misalignment to the laboratory schedule. A series of mixed model analyses were conducted with each discrete mood state as a dependent variable and fixed factors of Day (Baseline 1, Baseline 2, Sleep Deprivation) and Sex. All models specified a random effect of subject ID and were fully saturated, including all main and interaction effects. Post hoc tests were conducted using Least Significant Differences (LSD). Inferential statistics are shown in Table 1. Figures showing the effect of day and sex on each mood states are presented in Figure 2. Results of mixed model analyses revealed depressed mood, anxiety, anger, confusion, fatigue and energy significantly worsened following one night without sleep when compared to baseline days. There was a main effect of sex on anxiety, with females reporting more anxiety overall than the males. Post hoc tests examining the significant interaction between day and sex on anxiety revealed that sleep deprivation only heightened anxiety in females, with sleep deprivation having no significant effect on self-reported 11 Page 11 of 23
anxiety in males. A similar pattern of results was found for depressed mood, with only females reporting significant increases in depressed mood following sleep deprivation, but not males. It is important to note, however, that the increase in depressed mood in males from baseline day 2 to sleep deprivation day approached significance (p=.053). Lastly, while confusion was significantly higher following sleep deprivation than at either baseline day 1 or 2 for both males and females, the magnitude of the change from baseline to sleep deprivation was greater in females.
12 Page 12 of 23
13 Page 13 of 23
The relative impact of sleep loss on mood across the 36 hour period of continual wakefulness is displayed in Figure 3, which shows that while mood states worsen across the biological night, as expected, they do not recover the following day. Mood scores worsened around the time of the circadian nadir (0500h test bout) and then remained similarly low across the day. To compare the magnitude of effect between mood states a mixed model analysis was conducted with change scores from each mood state between baseline day 2 and sleep deprivation day (as displayed in Figure 3) as the dependent variable and mood a fixed factor. The model specified a random effect of subject ID. Results showed a significant main effect of mood on change scores, F(5,425)=87.69, p Vigour (19%) > Confusion (12%) > Anxiety (7%) > Depressed mood (3%), Anger (2%), all p
S1389-9457(15)00698-X http://dx.doi.org/doi:10.1016/j.sleep.2015.03.007 SLEEP 2728
To appear in:
Sleep Medicine
Received date: Revised date: Accepted date:
3-11-2014 26-3-2015 27-3-2015
Please cite this article as: Michelle A. Short, Mia Louca, Sleep deprivation leads to mood deficits in healthy adolescents, Sleep Medicine (2015), http://dx.doi.org/doi:10.1016/j.sleep.2015.03.007. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Sleep deprivation leads to mood deficits in healthy adolescents
Michelle A. Short, PhD & Mia Louca, BPsycSci(Hons) Centre for Sleep Research University of South Australia
This study was performed at the Centre for Sleep Research, University of South Australia and funded by a Divisional Research Performance Grant from the University of South Australia. The authors have no conflicts of interest to disclose.
Corresponding author: Michelle A. Short Centre for Sleep Research GPO Box 2471 Adelaide SA 5001 PH: +61 8 8302 1966 Email: [email protected] 1 Page 1 of 23
Highlights
The effect of sleep deprivation on mood was examined in healthy adolescents Sleep loss causally affected all discrete mood states Males and females showed differential vulnerability to sleep loss Discrete mood states differed in their sensitivity to sleep loss Abstract
Study Objectives: To investigate the effects of 36 hours of sleep deprivation on the discrete mood states of anger, depression, anxiety, confusion, fatigue and vigour in healthy adolescents. Method: Twelve, healthy adolescent good sleepers (6 male), aged 14 to 18 years (M=16.17, SD=0.83), spent three consecutive nights in the sleep laboratory of the Centre for Sleep Research: two baseline nights with 10h sleep opportunities and one night of total sleep deprivation. Every two hours during wake they completed the Profile of Mood States – Short Form. Mood across two baseline days was compared to mood at the same clock time (0900h to 1900h) following one night without sleep. Results: The subscales of depression, anger, confusion, anxiety, vigour and fatigue were compared across days. All mood states significantly worsened following one night without sleep. Females showed a greater vulnerability to mood deficits following sleep loss, with greater depressed mood and anxiety following sleep deprivation only witnessed among female participants. While males and females both reported more confusion following sleep deprivation, the magnitude of this effect was greater for females. Conclusions: This study provides empirical support for the notion that sleep loss can causally affect mood states in healthy adolescents, with females having heightened vulnerability. Understanding the detrimental effects of insufficient sleep during adolescence is important as it is a stage where sleep loss and mood dysregulation are highly prevalent. These findings escalate the importance of promoting sleep for the wellbeing of adolescents at this critical life phase.
KEYWORDS: Adolescence, Anxiety, Depressed Mood, Anger, Fatigue, Sleep, Affect, Emotion.
2 Page 2 of 23
INTRODUCTION Sleep deprivation has been linked to serious changes in mood states and their regulation (Haack & Mullington, 2005). While experimental studies have begun to elucidate the relationship between sleep and mood in adults, this relationship is underresearched in adolescents. Indeed, the lack of experimental research examining the impact of sleep on mood in adolescents has been identified as a critical gap in the literature (Talbot, McGlinchey, Kaplan, Dahl, & Harvey, 2010). To address this gap, present study examines the effect of 36 hours of sleep deprivation on the discrete mood states of depression, anxiety, anger, fatigue, vigour and confusion in an adolescent sample. Evidence for the relationship between sleep disorders and mood disorders is strong in adults and adolescents (Baglioni et al., 2011; Harvey, 2008; Lovato & Gradisar, 2014; Taylor, Lichstein, Durrence, Reidel, & Bush, 2005). Indeed, sleep problems are one of the diagnostic criteria for major depressive disorder. However, it is unclear to what degree sleep loss can perturb mood in healthy adolescents. Establishing a causal link between sleep and mood in healthy populations would provide important support for the causal relationship from insufficient sleep to mood deficits, as well as highlighting the importance of preventative public health strategies targeting sleep and well-being in this age group. As mood disorders are one of the leading causes of morbidity and mortality in adolescents (Forbes & Dahl, 2005), this is an important and timely area of research. Previous work examining sleep deprivation in adults has elucidated a range of functional deficits associated with sleep, as well as potential biological mechanisms underpinning them. Adult research has shown mood deficits in response to both sleep deprivation and sleep restriction, including decreased vigour and increased confusion and fatigue (Dinges et al., 1997; C. L. Drake et al., 2001). Some studies have also found
3 Page 3 of 23
increased anger (Haack & Mullington, 2005), anxiety (Brendel et al., 1990; Reynolds, Kupfer, Hoch, & Stack, 1986) and depressed mood (Scott, McNaughton, & Polman, 2006; Selvi, Gulec, Agargun, & Besiroglu, 2007) in response to sleep loss, however these mood states have not been shown to consistently worsen (Dinges et al., 1997; C. L. Drake et al., 2001). While vigour, confusion and fatigue worsen with moderate levels of sleep restriction, mood states, such as depressed mood, anger and anxiety, tend to worsen with more severe sleep restriction, and in particular, following total sleep deprivation (Babson, Trainor, Feldner, & Blumenthal, 2010). This suggests a differential sensitivity to sleep loss between mood states. In addition, there may be a dose-dependent relationship between sleep loss and mood deficits, with greater sleep loss predicting greater mood deficits. This is consistent with the dose-response relationships between sleep duration and other functional outcomes (Van Dongen, Maislin, Mullington, & Dinges, 2003). As such, the present study utilises a condition of total sleep deprivation, rather than sleep restriction, to operationalise sleep loss in order to best establish a causal relationship between sleep loss and a range of discrete mood states. In addition, the sensitivity of different mood states to the same degree of sleep loss will be examined. Despite many excellent studies in adults, there are several reasons why these findings cannot be generalised to adolescents. Adolescents differ from adults in terms of brain maturation, sleep homeostatic and circadian system function, and psychosocial milieu (Crowley, Acebo, & Carskadon, 2007; Dahl, 2004; Jenni, Achermann, & Carskadon, 2005). Adolescents are argued to have fewer resources to cope with the affective challenges they face because of their developmental stage (Talbot et al., 2010). Sleep has shown to have a role in regulating emotional brain-states, with the amygdala implicated in the processing of emotionally salient information (Yoo, Gujar, Hu, Jolesz, & Walker, 2007). In an fMRI study, Yoo and colleagues (Yoo et al., 2007) found an 4 Page 4 of 23
amplified hyper-limbic response by the amygdala to aversive emotional stimuli under conditions of sleep deprivationand reduced functional connectivity between the amygdala and the medial prefrontal cortex (mPFC). The mPFC has an inhibitory function over the amygdala, meaning that heightened emotional reactivity related to amygdala function is not attenuated to the same degree by the mPFC during conditions of sleep loss. As the prefrontal cortex (PFC) is still developing in adolescents (Talbot et al., 2010), mood deficits arising from sleep loss may be amplified. While experimental research into the relationship between sleep and mood in adolescents is limited, several excellent studies have begun to explore this relationship in healthy populations (Baum et al., 2014; Dagys et al., 2012; Dewald-Kaufmann, Oort, & Meijer, 2014; McGlinchey et al., 2011; Talbot et al., 2010). Talbot and colleagues (Talbot et al., 2010) restricted sleep to 6.5 hours for the first night and 2 hours for the second night and then compared mood (using the Positive Affect and Negative Affect Schedule for Children, or PANAS-C) to a rested condition in which participants slept between 7 and 8 hours per night for two nights. This effect of sleep loss was investigated across early adolescents, mid adolescents and adults. It was found that while adolescents reported less positive affect when sleep restricted compared to when rested, negative affect did not change. Adolescents also reported higher levels of anxiety following a catastrophizing ask and rated the likelihood of the catastrophes much higher when sleep deprived (Talbot et al., 2010). A study by Baum and colleagues (Baum et al., 2014) found changes to both positive and negative mood states following sleep loss. The authors conducted a homebased experimental sleep restriction protocol in a sample of 50 healthy adolescents aged 14 to 17 years. Participants completed one week of baseline sleep, followed by 5 nights of 6.5 hours’ time in bed (TIB) night and 5 nights of 10 hours’ TIB in a counterbalanced order, 5 Page 5 of 23
with one weekend as a washout period in between conditions. Participants reported significantly more anxiety, anger, confusion and fatigue, and significantly less vigour on the Profile of Mood States during the sleep restriction period compared to the extended sleep week. Contrary to expectations, depressed mood did not change between conditions. While these adolescent sleep restriction studies cast some doubt over the ability of sleep loss to perturb depressed mood, another study using a sleep extension paradigm found that more sleep resulted in less depressed mood (Dewald-Kaufmann et al., 2014). Fifty-five adolescents were identified as having chronic sleep reduction and randomised them to either a sleep extension and sleep hygiene group or a control group. Time in bed was extended via earlier bedtimes over a two week period by one hour and monitored actigraphically. At the end of the two week period, those in the sleep extension group obtained more sleep and reported decreased depressed mood on the Children’s Depression Inventory . Unfortunately, other mood states were not assessed. While these studies are invaluable in elucidating possible causal relationships between sleep duration and mood, results have been inconsistent between studies and there are a number of methodological challenges yet to be addressed. For example, while Talbot and colleagues (Talbot et al., 2010) bought participants into the laboratory for the second night of sleep restriction, the remainder of the sleep restriction/extension studies in adolescents utilised at-home protocols. These have the advantage of providing greater ecological validity, but, as with any home-based sleep experiment, this comes at a possible loss to experimental control. Beebe et al., (2008) acknowledged a potential for nonadherence to study protocols within home-based sleep restriction studies due to the lack of behavioural monitoring of participants. In one adolescent sleep restriction study, 15% of adolescent participants reported non-compliance with the protocol (Jiang et al., 2011) and a number of participants dishonoured prohibitions against drinking caffeine or napping. 6 Page 6 of 23
Relying on self-reported adherence to protocols means that participants may under-report their napping or caffeine use (Beebe et al., 2008). Some studies also allowed limited consumption of caffeine (Baum et al., 2014; Dewald-Kaufmann et al., 2014). As caffeine can eliminate the mood and alertness deficits caused by sleep loss (Penetar et al., 1993), this may have minimised or even ameliorated sleep-related mood deficits. Finally, there is uncertainty over the degree to which confounding factors known to affect both mood and sleep, such as physical activity (Rothon et al., 2010) or exposure to bright light (Golden et al., 2005), are present. To address these methodological concerns the present study utilised a laboratory-based protocol. This study will add to our understanding of the effect of sleep loss on adolescent mood: (i) by establishing whether sleep loss can perturb a range of mood states in adolescents, including depressed mood, (ii) by establishing whether mood states differ in their sensitivity to sleep loss, and (iii) by establishing whether male and female adolescents differ in their mood response to sleep loss. Participant mood following one night of total sleep deprivation was compared to that across two baseline days following 10 hour sleep opportunities. It was hypothesised that there would be significantly increased depressed mood, anxiety, anger, fatigue and confusion and decreased energy following one night without sleep, when compared to two baseline periods when 10 hour sleep opportunities are given. Based on findings from adults sleep deprivation studies (Brendel et al., 1990; Dinges et al., 1997; C. L. Drake et al., 2001; Haack & Mullington, 2005; Reynolds et al., 1986; Scott et al., 2006; Selvi et al., 2007), it is hypothesised that mood changes in response to sleep loss will be significantly greater for the discrete mood states of confusion, vigour and fatigue than for anger, anxiety and depressed mood. Finally, there is some argument that mood deficits in response to sleep loss may be greater in females than males (Armitage & Hoffmann, 2001). However, 7 Page 7 of 23
due to the lack of previous research addressing sex differences across the range of mood states, these analyses were exploratory. METHOD Participants Participants included 12 adolescents (50% male) aged 14 to 18 years (M=16.17, SD=0.83) who attended secondary schools in South Australia. Participants were physically and psychologically healthy, as determined by parent- and self-reported surveys.. Participants who obtained less than 8 hours sleep per night, had average sleep onset latencies of more than 30 minutes per night, weekend bedtime delay of two hours or more or who were extreme morning or evening chronotypes (total scores ≤ 22 or ≥44) were excluded from the study, as were participants who scored of 16 or greater on the CES-D or received scores above “mild” on any of the subscales of the DASS-21. Participants did not use medication and were free of any medical, psychological and sleep disorders, as determined by parent and adolescent self-report questionnaires, a 7-day sleep diary and wrist actigraphy. Participants were either late- or post-pubertal (Tanner stages 4 or 5), as measured by the Pubertal Development Scale (Petersen, Crockett, Richards, & Boxer, 1988) and had normal or corrected-to-normal vision. Ethics approval was granted by the University of South Australia Human Research Ethics Committee and participants received an honorarium for their participation. Written, informed consent was obtained from each adolescent and a parent. Mood Measure The Profiles of Mood States Questionnaire-Short Form (POMS-SF) (Shacham, 1983) was used to measure mood. The POMS-SF consists of a list of 37 adjectives 8 Page 8 of 23
describing mood states that fall into one of six subscales: Anxiety (N = 6 items), Vigour (N = 6 items), Depressed mood (N = 8 items), Fatigue (N = 5 items), Anger (N = 7 items) and Confusion (N = 5 items) (Curran, 1995). Respondents rate the extent to which they are experiencing that mood state at that point in time on a 5-point Likert scale which ranges from 1 (not at all) to 5 (extremely). For each subscale, higher scores indicate greater presence of that mood. To enable comparison of relative impact of sleep loss on mood across the 36 hour period of continual wakefulness, and to account for differences in the number of items within each mood subscale, average scores were calculated for each subscale at each time point, including those across the night. Procedure The present study utilised a repeated measures design. The independent variables were day (baseline 1, baseline 2, sleep deprivation) and time (09:00h, 11:00h, 13:00h, 15:00h, 17:00h, 19:00h) and the dependent variables were Anxiety, Vigour, Depressed Mood, Fatigue, Anger and Confusion. Parents of participants underwent a telephone screening interview using The Sleep, Medical, Education and Family History Survey. Suitable participants were invited to attend an interview at the Centre for Sleep Research where the adolescents completed a screening battery Sleep, Medical, Education and Family History survey and a questionnaire battery containing the School Sleep Habits Survey (Wolfson et al., 2003), Centre for Epidemiological Studies Depression Scale (CES-D) (Radloff, 1977), Composite Morningness/Eveningness Scale (Smith, Reilly, & Midkiff, 1989), Depression Anxiety and Stress Scale (21 item; DASS-21) (Lovibond & Lovibond, 1995), and Pediatric Daytime Sleepiness Scale (C. Drake et al., 2003) and were given a one-week sleep diary 9 Page 9 of 23
and activity monitor to use over one typical school week. Three potential participants were excluded following screening due to long sleep onset latencies, heightened sleep variability or elevated anxiety. During the week prior to the study, participants maintained a regular sleep pattern, with bedtime at 2200h and wake time at 0700h (Roberts, Roberts, & Duong, 2009) in order to remove any sleep debt and ensure similar circadian entrainment between participants. Adherence was verified using sleep diaries, wrist actigraphy, and timestamped messages to the sleep laboratory morning and night. All participants who began the pre-study sleep schedule maintained adherence to the schedule and completed the study. Participants spent 3 consecutive nights in the Sleep Laboratory of the Centre for Sleep Research, from 1600h on day 1, until 2000h on day 4 (see Figure 1). The sleep laboratory is sound attenuated and light and temperature controlled. Light exposure was kept to less than 50 lux during all wake periods and less than 1 lux during sleep opportunities. Ambient temperature was 21°C (±1°C) at all times. Participants completed a neurobehavioural test battery (NTB) every two hours during wake. The NTB included a 10 minute psychomotor vigilance task, digit symbol substitution task, POMS-SF and the Karolinska Sleepiness Scale every two hours during wake period. NTBs began at 0900h to avoid the effects of sleep inertia on mood in the period immediately following wake. Results of the neurobehavioural measures are presented elsewhere (Louca & Short, 2014). On the first 2 baseline nights the 10-hour sleep opportunities were recorded using polysomnography (PSG), with a montage including frontal (F3/F4), central (C3,C4), and occipital (O1,O2) referenced against A1/A2, bilateral electrooculography, electromyography and electrocardiography. In between test bouts, participants could interact with fellow participants and research staff, watch DVDs, play card and board
10 Page 10 of 23
games and complete jigsaw puzzles. Continuous behavioural observation ensured that participants remained awake during all designated wake periods.
RESULTS
On average, participants obtained approximately 9 hours sleep during the two baseline sleep opportunities in the laboratory (X = 537.08 min, SD = 21.47), establishing the adequacy of the baseline nights to ensure participants were well-rested coming into the sleep deprivation night. Average sleep onset latencies of 25.60 minutes (SD = 15.58) per night indicated that participants were able to initiate sleep in the laboratory setting and were unlikely to be experiencing circadian misalignment to the laboratory schedule. A series of mixed model analyses were conducted with each discrete mood state as a dependent variable and fixed factors of Day (Baseline 1, Baseline 2, Sleep Deprivation) and Sex. All models specified a random effect of subject ID and were fully saturated, including all main and interaction effects. Post hoc tests were conducted using Least Significant Differences (LSD). Inferential statistics are shown in Table 1. Figures showing the effect of day and sex on each mood states are presented in Figure 2. Results of mixed model analyses revealed depressed mood, anxiety, anger, confusion, fatigue and energy significantly worsened following one night without sleep when compared to baseline days. There was a main effect of sex on anxiety, with females reporting more anxiety overall than the males. Post hoc tests examining the significant interaction between day and sex on anxiety revealed that sleep deprivation only heightened anxiety in females, with sleep deprivation having no significant effect on self-reported 11 Page 11 of 23
anxiety in males. A similar pattern of results was found for depressed mood, with only females reporting significant increases in depressed mood following sleep deprivation, but not males. It is important to note, however, that the increase in depressed mood in males from baseline day 2 to sleep deprivation day approached significance (p=.053). Lastly, while confusion was significantly higher following sleep deprivation than at either baseline day 1 or 2 for both males and females, the magnitude of the change from baseline to sleep deprivation was greater in females.
12 Page 12 of 23
13 Page 13 of 23
The relative impact of sleep loss on mood across the 36 hour period of continual wakefulness is displayed in Figure 3, which shows that while mood states worsen across the biological night, as expected, they do not recover the following day. Mood scores worsened around the time of the circadian nadir (0500h test bout) and then remained similarly low across the day. To compare the magnitude of effect between mood states a mixed model analysis was conducted with change scores from each mood state between baseline day 2 and sleep deprivation day (as displayed in Figure 3) as the dependent variable and mood a fixed factor. The model specified a random effect of subject ID. Results showed a significant main effect of mood on change scores, F(5,425)=87.69, p Vigour (19%) > Confusion (12%) > Anxiety (7%) > Depressed mood (3%), Anger (2%), all p
