Exploring the relationship between sleep and aggression in adolescents: A cross sectional study using the UK Millennium cohort study

Abstract

This study aimed to establish if a significant relationship exists between sleep and aggression in a large representative adolescent cohort and explores the impact of potential confounders. This cross-sectional secondary data analysis included 10,866 males and females aged 13–15 years, from the UK-based 2015 Millenium Cohort Study (sixth sweep). Independent variables included self-report measures of sleep duration and quality. The parent reported ‘Strengths and Difficulties Questionnaire’ conduct score measured aggression. Binary logistic regression examined independent associations between each sleep variable and aggression. Multiple regression models then adjusted for potential confounders: age, sex, socioeconomic status, arousal, and affect. Under 8 hours of sleep on average was significantly associated with aggression when age, sex and income were controlled (p = .008). This became insignificant following adjustment for both affect and arousal. Sleep quality remained significantly associated with aggression when all confounders were controlled: ‘sleep onset latency >30 minutes’ and ‘wakening at least a good bit of the time’ increased the odds of aggression by around 27.9% (p < .001) and 43.5% respectively (p < .001). A significant association exists between poor subjective sleep quality and heightened aggression in this cohort, when all our confounders are controlled, identifying sleep quality as a potential target in treating adolescent aggression.

Keywords

sleep aggression affect arousal adolescents hyperactivity

Introduction

Adolescent aggression negatively impacts individuals, the economy and society. Aggression is one of the most common reasons for adolescent psychiatric referrals and has been linked to reduced academic performance in teenagers (Granvik Saminathen, 2021; Pikard et al., 2018). Heightened adolescent aggression may also trigger verbal or physical assault, ranging from bullying to potentially serious violence, and can predict future crime (Friedman et al., 2021). These consequences of aggression increase pressure on health and social services. Youth violence cost around £1.3 billion in England and Wales during 2018, but more importantly, it is the fourth leading cause of death in those aged 10–29 years worldwide. (Billingham et al., 2020; World Health Organisation, 2020).

Therefore, prevention of adolescent aggression should be a global priority. Multiple factors, including social circumstances and hormonal changes, contribute to high aggression rates in adolescents (Najman et al., 2009). Further understanding of factors impacting adolescent aggression is needed to inform treatment and develop preventative public health strategies. Poor sleep has been associated with increased adolescent aggression (El-Sheikh et al., 2019; Van Veen et al., 2021).

Sleep and adolescent aggression

Recently, a robust systematic review and meta-analysis found a consistent association between poor sleep quality and increased aggression across all ages, and similar results have been seen across numerous observational cohort studies (Van Veen et al., 2021).

Many psychosocial, environmental, and biological factors, including circadian changes, predispose teenagers to sleep problems (Crowley et al., 2007; Suni, 2022a). This poor sleep, together with social and neurodevelopmental changes specific to adolescence, such as prefrontal cortex (PFC) and default mode network (DMN) development, likely increase the susceptibility of this age group to aggression (Arain et al., 2013).

Despite this potential vulnerability, research on the relationship between sleep and aggression in adolescents is lacking. Observational studies have found an association between poor sleep and increased adolescent aggression, however there are mixed results and limitations throughout this evidence base. Firstly, many published studies have reported findings from specific populations, including juvenile offenders, which are unlikely to produce generalisable results (Ireland & Culpin, 2006). More observational evidence from large representative adolescent cohorts is needed. Secondly, methodological inconsistencies between studies have produced conflicting results and limit the comparability of studies. For example, disagreement has been observed between objective sleep measures, such as actigraphy, and subjective sleep measures, including sleep diaries (Tremaine et al., 2010).

Varying definitions of sleep and aggression also influence results. Sleep is often defined as sleep duration, sleep quality, or both (El-Sheikh et al., 2019). Some studies support a relationship between short sleep duration and increased aggression, but more recent evidence argues that sleep quality is more strongly associated with aggression than sleep duration (Clinkinbeard et al., 2011; Connelly et al., 2021; Vermeulen et al., 2021).

The general aggression model (GAM)

Another limitation of many studies is the absence of a model explaining the relationship between sleep and adolescent aggression. The general aggression model is a well-known theoretical model integrating cognitive, psychosocial, and biological factors which influence the pathways leading to aggression (Allen & Anderson, 2017). ‘Distal processes’ include biological mechanisms (e.g., hormone imbalances) and environmental modifiers (e.g., social deprivation), which feed into the ‘proximate causes’ that lead to aggressive episodes. Proximate inputs involve person and situation factors (e.g., social stress). These alter affect, cognition, and arousal, which influence decision making and re-appraisal abilities, leading to aggressive behaviour (Allen & Anderson, 2017). Aggressive behaviour can then feedback to cause further behavioural change through altered personality and knowledge structures (Allen & Anderson, 2017).

Studies have investigated sleep as a potential biological modifier, and three psychological routes have been hypothesised from poor sleep to aggression through the general aggression model. These involve increased perceived hostility through the ‘cognitive pathway’, reduced impulse inhibition through the ‘response-control’ pathway, and increased negative affect through the ‘affective pathway’ (Krizan & Herlache, 2016).

This exploratory research addresses some limitations in existing evidence, by using a large population-based sample, including measures of both sleep quality and quantity, and using a theoretical model to inform the research question, choice of potential confounders and interpretation of findings.

The aim of this study was not to test the general aggression model, instead this was an exploratory study which examined the following research questions:

• Is there a significant relationship between sleep and aggression in adolescents?

• Does the relationship between sleep and aggression in adolescents persist after controlling for the confounding effects of age, sex, socioeconomic status, affect and arousal?

Material and methods

Study population

The Millenium Cohort Study (MCS) is a multi-disciplinary, nationally representative prospective birth cohort study which began with over 18,000 children aged 9 months old who were born in 2000-2001 across the United Kingdom. Subsequent data collection occurred around ages 3, 5, 7, 11, and 14 years, primarily through questionnaires (Fitzsimons et al., 2020). The stratified, clustered random sample design resulted in oversampling from disadvantaged areas and areas with high ethnic minority populations (Fitzsimons et al., 2020). Our sample from January 2015 to March 2016, during the sixth wave of the MCS (MCS6), involved 10,866 adolescents aged 13–15 years. For convenience, our sample included only the first participant from each family, and those who provided conclusive answers to the following domains: sleeping habits, income, Strengths and Difficulties Questionnaire (SDQ) conduct problems score, SDQ emotional symptoms score, and SDQ hyperactivity/inattention score. Those with inconclusive responses, or missing data for any of these variables were excluded from analysis (Supplementary Material Figure 1). The MCS6 was approved by the Research Ethics Committee London - Central.

Sleep characteristics

Supplementary Material Table 1 summarises the MSC6 self-report sleeping questions. Sleep onset and wake times were used to calculate approximate average nightly sleep duration for each cohort member, using a 5:2 ratio of school nights to non-school nights. These values were collapsed into four groups: ‘<8 hours’, ‘8–9 hours’, ‘>9–10 hours’ and ‘>10 hours’.

Under 8 hours was considered short sleep, as adolescents are recommended 8–10 hours of sleep per night (Suni, 2022b). Sleep onset latency and night-time awakening are typical indicators of sleep quality (Suni, 2022b). Categories: ‘30 minutes or less’ and ‘over 30 minutes’ were used for sleep onset latency, with over 30 minutes indicating poor-quality sleep. To simplify analyses, two categories were also created for night-time awakening: ‘At least a good bit of the time’ which indicated poor sleep quality, and ‘Some, little or none of the time’.

Strengths and difficulties questionnaire

MCS6 included the SDQ and impact supplement for parents of 4–17 year-olds. This widely used, validated behavioural questionnaire designed for the general population was suitable for our large, non-clinical sample. The SDQ consists of 25 items divided among 5 scales (Supplementary material Table 2). Parents responded, ‘not true’, ‘somewhat true’ or ‘certainly true’, to statements on the form, creating a numerical score for each subscale. The ‘conduct problems’ scale was used to investigate aggressive behaviour, as this has been used as a proxy measure of externalising behaviour or aggression in epidemiological research examining the relationships between sleep and aggression (Van Veen et al., 2021). ‘Emotional symptoms’ and ‘hyperactivity/inattention’ SDQ scores were used to explore the potential impact of affect and arousal respectively as guided by the general aggression model.

In practice, SDQ subscale scores are categorised into four groups based on defined numerical cut-offs (2020). To simplify analyses, binary variables were created for each of the SDQ measures based on these recommended groupings (Supplementary Material Table 2). These included the recommended ‘close to average’ group, which contains around 80% of a representative population. The remaining three recommended groups: ‘above average’, ‘high’ and ‘very high’ were merged into a single category labelled as ‘above average’, which is expected to contain around 20% of a representative population (Supplementary Material Table 2). Therefore, aggression was indicated by an above average conduct score between 3-10, hyper-arousal was indicated by an above average hyperactivity score between 6–10, and altered affect was indicated by an above average SDQ emotional symptoms score between 4–10.

Confounders

Potential confounders were based on the general aggression model and evidence from existing research. These included sex at birth and age as potential biological modifiers, socioeconomic status as a potential environmental modifier, and affect and arousal as possible routes in the relationship between sleep and aggression (Allen & Anderson, 2017). ‘Age at last birthday’ measured age. We also adjusted our analyses for sex at birth, as higher direct aggression levels are commonly reported in males (Lansford et al., 2012). Annual income quintiles, equivalised through a modified Organisation for Economic Co-operation and Development (OECD) scale, were used to measure socioeconomic status. These quintiles are weighted using the overall weights in the MCS6, and for the UK, ensuring this is a representative indicator that is comparable across UK countries. As mentioned, SDQ ‘emotional symptoms’ subscale, was used to adjust for affect, and SDQ ‘hyperactivity/inattention’ subscale was used to control arousal (Supplementary Material Table 2).

Statistical methods

Binary logistic regression models were created, and each model was repeated three times, with above average SDQ conduct problems score as the outcome, and a different sleep parameter (sleep duration category, sleep onset latency or night-time awakening) as the dependent variable each time. This allowed each of the three sleep parameters to be investigated independently from one another, and therefore detect if any association with aggression differed between the sleep duration and sleep quality indicators. The initial model controlled no covariables. Model 1 was adjusted for age and sex. Model 2 was adjusted for age, sex and OECD-equivalised income quintiles. Model 3 was fully adjusted for our covariables: age, sex, socioeconomic status (OECD-equivalised income), affect (SDQ emotional symptoms) and arousal (SDQ hyperactivity). Odds ratios (ORs) and 95% confidence intervals (CIs) were calculated using SPSS software (version 28) and p-values<.05 were considered statistically significant.

Hosmer and Lemeshow goodness of fit test examines the fit of the models, and Nagelkerke R² estimates the strength of the relationship between predictors and outcomes. The Omnibus test indicates whether a more complex model is a statistically significant stronger relationship than a simpler model (Supplementary Material Table 3).

Results

Supplementary Material Figure 1 outlines how the sample was obtained, and Supplementary Material Table 4 summarises key population characteristics.

The mean SDQ conduct problems score was 1.38, which is around average in the general population (2020). The mean nightly sleep duration in this sample was 9.2 hours, and 9% reported less than 8 hours of sleep per night on average.

Around half of this sample were males (49.8%) and around half were females (50.2%). Significant sex differences were seen across some of the variables (Supplementary Material Table 4). Girls generally reported a shorter sleep duration, longer sleep onset latency and more night-time awakenings than boys. Girls also had significantly higher SDQ emotional symptoms scores, whilst boys had significantly higher SDQ hyperactivity scores than girls.

The initial binary logistic regression model found significant associations between each of the individual sleep measures and above average conduct problems (Table 1). When no confounders were controlled, sleep duration ‘<8 hours’ was significantly associated with increased odds of above average conduct problems, (OR = 1.357, 95% CI: 1.120-1.644, p = .002) compared to ‘>10 hours’, whilst sleep duration ‘>9–10 hours’ was significantly associated with reduced odds of above average conduct problems (OR = .839, 95% CI: .723-.972, p = .019) compared to ‘>10 hours’.

Table 1.

Each Model From the Multiple Regression Analysis Showing the Association of Each Individual Sleep Parameter With Above Average SDQ Conduct Problems Score: Odds Ratios (ORs), Confidence Intervals (CIs) and P-values Shown.

Step of the Multiple Regression Model	Sleep Parameter		Above average SDQ Conduct Score
			OR (CI)	p value
Initial Model	Sleep duration^b	<8 hours	1.357 (1.120–1.644)	.002^a
		8-9 hours	1.015 (.874–1.178)	.848
		>9-10 hours	.839 (.723–.972)	.019^a
		>10 hours	(Reference category)
	Sleep Onset Latency (in previous 4 weeks)^c	30 minutes or more	1.432 (1.298–1.580)	<.001^a
	Night-time awakening (in previous 4 weeks)^d	At least a good bit of the time	1.994 (1.787–2.225)	<.001^a

Model 1 (Adjusted for age and sex)	Sleep duration^b	<8 hours	1.364 (1.126–1.654)	.002^a
		8-9 hours	1.017 (.876–1.181)	.823
		>9-10 hours	.840 (.725–.974)	.021^a
		>10 hours	(Reference category)
	Sleep Onset Latency (in previous 4 weeks)^c	30 minutes or more	1.443 (1.307–1.593)	<.001^a
	Night-time awakening (in previous 4 weeks)^d	At least a good bit of the time	2.041 (1.827–2.281)	<.001^a

Model 2 (Adjusted for age, sex and OECD-equivalised income quintiles)	Sleep duration^b	<8 hours	1.307 (1.074–1.590)	.008^a
		8-9 hours	1.089 (.935–1.268)	.275
		>9-10 hours	.896 (.770–1.041)	.152
		>10 hours	(Reference category)
	Sleep Onset Latency (in previous 4 weeks)^c	30 minutes or more	1.446 (1.307–1.600)	<.001^a
	Night-time awakening (in previous 4 weeks)^d	At least a good bit of the time	1.790 (1.598–2.005)	<.001^a

Model 3 (Adjusted for age, sex, OECD-equivalised income, SDQ emotional symptoms and SDQ hyperactivity)	Sleep duration^b	<8 hours	1.184 (.953–1.472)	.128
		8-9 hours	1.166 (.987–1.379)	.072
		>9-10 hours	.949 (.805–1.120)	.539
		>10 hours	(Reference category)
	Sleep Onset Latency (in previous 4 weeks)^c	30 minutes or more	1.279 (1.144–1.429)	<.001^a
	Night-time awakening (in previous 4 weeks)^d	At least a good bit of the time	1.435 (1.265–1.627)	<.001^a

^aRepresents statistically significant result (p < .05).

^bORs measure the increase (>1) or decrease (<1) in the odds of having an above average SDQ conduct problems score for those reporting ‘<8 hours’, ‘8–9 hours’ or ‘>9–10 hours’ of sleep respectively, compared to ‘>10 hours’.

^cORs measure the increase (>1) or decrease (<1) in the odds of having an above average SDQ conduct problems score for those reporting a sleep onset latency >30 minutes compared to those reporting a sleep onset latency ≤30 minutes.

^dORs measure the increase (>1) or decrease (<1) in the odds of having an above average SDQ conduct problems score for those who waken ‘at least a good bit of the time’ at night, compared to those who waken ‘some, little, or none of the time’ at night.

Controlling for age and sex only slightly altered the odds ratios and p-values and did not change the significance of the results (Table 1). Further controlling for income had minimal impact, although Table 1 shows that when age, sex and income were controlled, the negative association between ‘>9–10 hours’ of sleep and above average conduct problems was no longer significant compared to ‘>10 hours’. The indicators of short sleep and those of poor sleep quality remained significantly associated with increased conduct problems.

Table 1 shows that the significant association between ‘<8 hours’ sleep duration and above average conduct problems disappeared when the regression model was further adjusted for emotional symptoms and hyperactivity in Model 3 (p = .128). Indicators of poor sleep quality remained significantly associated with above average conduct problems in this final, fully adjusted multivariate model. The odds of having above average conduct problems were estimated to be around 27.9% higher in those reporting a sleep onset latency greater than 30 minutes, compared to under 30 minutes (OR = 1.279, 95% CI: 1.144-1.429, p < .001).

Moreover, the odds of having above average conduct problems were estimated to be around 43.5% higher in those who woke at least a good bit of the time at night compared to those who woke some, little, or none of the time (OR = 1.435, 95% CI: 1.265-1.627, p < .001).

Supplementary Table 3 outlines the goodness of fit and the explanatory strength of the models. For each sleep parameter, from the initial model to model 3 the deviance decreases significantly, and the R-square increases, indicating that the addition of covariables created a final model which explained a greater proportion of the variance in the relationship between sleep and aggression than that of the initial model. The Hosmer–Lemeshow test indicates that models 1 and 2 are a good fit for each sleep parameter, but model 3 may not be a good fit, though the excessive power of this test in large sample sizes similar to this study can lead to very small departures from the model being labelled significant (Kramer & Zimmerman, 2007).

Discussion

This exploratory study reports several notable findings. Firstly, both shorter and poorer quality sleep were associated with increased adolescent aggression as measured by the SDQ conduct score when age and sex were controlled. This finding agrees with current literature. Secondly, income had a minimal impact on these associations, which was unexpected, as low socioeconomic status has been identified as a vulnerability factor for increased adolescent rule-breaking behaviour following poor sleep (El-Shiekh et al., 2019). Thirdly, emotional symptoms and hyperactivity appear to act as confounders in this relationship and were responsible for a significant proportion of the association between short sleep duration and increased adolescent aggression, as supported by the theoretical roles of affect and arousal in the general aggression model (Allen & Anderson, 2017). Finally, the key finding was a significant association between poor subjective sleep quality, but not duration, and increased adolescent aggression when confounders were fully adjusted, which aligns with a growing evidence base suggesting that sleep quality is more strongly related to adolescent aggression than sleep duration (Connolly et al., 2021; Vermeulen et al., 2021).

Is sleep quality more important than sleep duration in adolescent aggression?

Sleep quality is considered to be more important than sleep duration across both physical and psychological health outcomes (Pilcher et al., 1997). Moreover, recent observational studies in adolescents found that poor subjective sleep quality, rather than sleep duration, was significantly associated with increased delinquency and worse psychological functioning when shared familial and environmental factors were controlled (Connolly et al., 2021; Vermeulen et al., 2021). Evidently, longer sleep is not always better. For example, cognitive behavioural therapy for insomnia (CBT-I) treatment involves reducing time spent in bed to improve sleep quality, and this has been shown to reduce oppositional symptoms in adolescents (de Bruin et al., 2018).

It is unclear why sleep quality may be more strongly related to adolescent aggression than sleep duration. Firstly, sleep quality has been recommended as a better measure of sleep than duration (Kohyama, 2021; Pilcher et al., 1997). Secondly, considering the vast number of possible covariables and the uncertainty surrounding underlying mechanisms, it is possible that uncontrolled confounders have increased the apparent strength of the relationship between sleep quality and aggression across the literature. One study associated poor subjective sleep quality, but not duration, with increased trait hostility in adults, suggesting that poor quality sleep may contribute to aggression through the cognitive pathway of the general aggression model (Allen & Anderson, 2017; Freitag et al., 2017).

However, studies suggest that both sleep quality and duration can affect DMN functioning, and a recent cross-sectional study found that interactions between sleep quality and duration were linked to increased rule breaking behaviour in a relatively small sample of 235 adolescents (Dai et al., 2015; El-Sheikh et al., 2019; Tashjian et al., 2018). These mixed results suggest that while poor sleep quality appears to have a stronger relationship with increased adolescent aggression than short sleep duration, these factors are likely intertwined. Therefore, despite our findings, sleep duration should not be discounted as an important risk factor for aggression in adolescents.

Nature of the relationship between sleep and adolescent aggression

What do we know Regarding Causation and Direction?

Neuroimaging studies in adolescents have linked short and disrupted sleep to abnormalities in brain structures involved in emotional control and aggression, including the DMN, the salience network and the PFC (Strenziok et al., 2011; Sung et al., 2020; Tashjian et al., 2018). Additionally, poor sleep has been associated with neurobiological abnormalities involving genes, the hypothalamic pituitary axis (HPA), and serotonin, which may increase aggression (Madrid-Valero et al., 2019; Meerlo et al., 2008; Roman et al., 2005). This experimental evidence, coupled with observations of reduced aggression following sleep treatment, suggest that poor sleep may reversibly cause aggression (Haynes et al., 2006). The general aggression model also indicates a potential bidirectional relationship between sleep and aggression, however, evidence on this bidirectional relationship is limited and conflicting and it was not within the scope of this study to investigate this. Unfortunately, due to the numerous confounders, complexity of underlying mechanisms and the design of this correlational, exploratory study, we are unable to determine the exact underlying mechanisms or direction of any interactions linking sleep, affect, arousal and adolescent aggression.

Affect and arousal as potential confounders in this study

Experimental and neuroimaging studies have found that poor sleep alters affect and arousal (Meerlo et al., 2008). For example, chronic partial sleep deprivation has resulted in reduced serotonin receptor sensitivity in rats, a common finding in depression (Roman et al., 2005). Additionally, sleep disruption and deprivation are associated with altered regulation of the HPA, therefore increasing arousal (Meerlo et al., 2008). Altered affect and arousal following poor sleep are thought to impact aggressive behaviour, which may then further disrupt sleep and psychological functioning, according to the general aggression model.

Abnormalities in particular brain regions involved in control of affect, arousal, and emotional regulation, such as the DMN, have been observed across adolescent populations reporting poor sleep, depression, or attention deficit hyperactivity disorder (ADHD) (Ho et al., 2015; Sung et al., 2020; Van Rooij et al., 2015). Considering these overlapping neurobiological findings, it is unsurprising that youths with heightened aggression, such as those with conduct disorder, often display co-morbid depression, ADHD, and/or poor sleep (Baker, 2013). These associations, coupled with the indirect effect of emotional symptoms and hyperactivity on the association between short sleep and adolescent aggression in this study, and interpreted alongside the general aggression model, indicate that negative affect and hyper-arousal may amplify the relationship between poor sleep (especially short sleep) and aggression in adolescents.

However, many people with hyperactivity or altered affect may experience poor sleep in the absence of aggression. Therefore, more experimental evidence is needed to elucidate the causal relationship between these factors and identify other influencing elements.

Socioeconomic status as a potential confounder in our study

El-Sheikh, et al. (2019) found that socioeconomic status interacted with sleep to predict adolescent rule breaking behaviour, and lower socioeconomic status acted as a vulnerability factor in this relationship. There are many possible reasons for this, including the effect of photic (e.g., use of mobile phones at night) and non-photic (e.g., overcrowding, social factors) zeitgebers on circadian rhythm, which are linked to socioeconomic status (Mistlberger & Skene, 2004). The surprisingly weak effect of socioeconomic status in this study may be partly attributable to the use of income rather than a more comprehensive indicator, such as the index of multiple deprivation (IMD). However, the IMD is not standardised across UK countries and therefore was not comparable across this sample.

Strengths and limitations

This large, non-clinical sample was from a nationally representative cohort study, increasing the generalisability of the results. Furthermore, this study investigates a particularly vulnerable age group, as adolescent aggression peaks around age 14, and the choice of covariables was guided by the general aggression model: a widely used theoretical model (Allen & Anderson, 2017; Karriker-Jaffe et al., 2008). Considering the conflicting literature regarding sleep quality and duration, multiple sleep parameters were used, covering both variables separately (Connolly et al., 2021; El-Sheikh et al., 2019; Vermeulen et al., 2021).

These multiple sleep parameters, and a conduct score which covered aspects of both physical and social aggression, resulted in a broad overview of sleep and aggression in adolescents.

Moreover, the parent SDQ form is suitable for our population-based sample, has sound psychometric properties and has been used across many cultures (Stone et al., 2015; Woerner et al., 2004). Finally, to enhance our study quality, statistical methods were discussed with a statistician prior to analysis.

However, sleep variables were limited to subjective, self-report questions. Although subjective sleep measures are important, they have been associated with overestimation of adolescent sleep duration when compared with objective measures such as actigraphy, which could have contributed to the insignificant relationship between sleep duration and aggression in the final regression model in Table 1 (Barker et al., 2016; Tremaine et al., 2010). Another limitation was the exclusion of around 8.4% of participants, partly due to missing data and partly due to ease of analysis. If given more time, multiple imputation could be explored to overcome any effect of missing data. Also, more time would have allowed for incorporation of specific MCS survey weights in analyses, which better account for the unique study design, although controlling income may have partly accounted for oversampling from disadvantaged areas within MCS6 (Fitzsimons et al., 2020). Additionally, current literature is limited by the inability to control many possible confounders as demonstrated by the complexity of the general aggression model, and interpretation of results must consider the potential impact of uncontrolled confounders.

Recommendations for future research and clinical practice

This study highlights the significance of subjective sleep quality in relation to adolescent aggression, and the importance of controlling psychosocial factors, especially affect and arousal, in future observational studies. This study could be extended by further exploring the relationship between sleep quality and adolescent aggression while controlling other potential confounders related to the general aggression model, such as cognitive factors (Allen & Anderson, 2017). Furthermore, longitudinal observational research and neuroimaging studies in large representative adolescent cohorts are needed to uncover causation and the direction of this relationship.

These results suggest that sleep quality should be prioritised in public health policies such as sleep hygiene education in schools. Few sleep intervention studies directly assess aggression, but one behavioural sleep intervention programme found decreased aggression in substance-abusing adolescents after six weeks (Haynes et al., 2006). Additionally, CBT for insomnia has reduced oppositional, ADHD and anxiety symptoms in adolescents for up to 12 months after treatment (de Bruin et al., 2018). Clinically, it may be useful to incorporate behavioural sleep-improvement techniques within established aggression treatments, such as cognitive behavioural therapy.

Conclusion

This cross-sectional study in a normative sample of adolescents provides robust evidence of a significant association between subjective poor sleep quality and increased aggression, when controlling for important environmental, biological, and psychosocial factors. Therefore, sleep quality is a potential target for preventing and treating adolescent aggression.

Conversely, sleep duration was not significantly associated with adolescent aggression when emotional symptoms and hyperactivity were controlled in the final model. This finding suggests that affect and arousal are important confounders in the relationship between sleep duration and adolescent aggression.

More research in large, representative, adolescent cohorts is needed to explain why sleep quality appears to be more strongly associated with adolescent aggression than sleep duration, and to identify potential underlying confounders in this relationship. Experimental and neuroimaging studies are required to determine the cause-effect relationship between sleep and adolescent aggression.

Supplemental Material

Supplemental material - Exploring the relationship between sleep and aggression in adolescents: A cross sectional study using the UK Millennium cohort study

Supplemental Material for Exploring the relationship between sleep and aggression in adolescents: A cross sectional study using the UK Millennium cohort study by Caoimhe McCaffery, John McClure, Sukhmeet Singh, and Craig A Melville in Clinical Child Psychology and Psychiatry

Footnotes

Acknowledgments

I would like to thank Professor Craig Melville for his valued support and supervision of this research. I would also like to thank Dr. Sukhmeet Singh and Dr. John McClure for their advice and guidance.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Ethical Statement

ORCID iD

Caoimhe McCaffery

Craig A Melville

Data Availability Statement

The data used in this study is derived from the sixth wave of the Millenium Cohort study. This data is openly available through the UK Data Service at: .

Supplemental Material

Supplemental material for this article is available online.

Author biographies

Caoimhe McCaffrey is a final year medical student at the University of Glasgow. This research was completed as part of her intercalated BSc degree in clinical medicine. Caoimhe has a strong interest in paediatric healthcare, and has presented her work at the 2023 Royal College of Paediatrics and Child Health conference.

John McClure, PhD is a Lecturer in Statistics in the School of Cardiovascular and Metabolic Health at the University of Glasgow

Sukhmeet Singh is a Higher Psychiatry Trainee in Intellectual Disabilities and Child and Adolescent Psychiatry working in the West of Scotland. His research interests primarily concern the care and treatment of children with intellectual disabilities.

Craig A Melville is a consultant psychiatrist and Professor of Neurodevelopmental Psychiatry at the University of Glasgow. He has a particular interest in the mental health of people with neurodevelopmental conditions and supporting the career development of early career researchers

References

Allen

J. J.

Anderson

C. A.

(2017). General Aggression Model. The international encyclopedia of media effects. Wiley. p. 1–15. https://www.craiganderson.org/wp-content/uploads/caa/abstracts/2015-2019/17AA.pdf (Accessed 28/12/2021).

Arain

Haque

Johal

Mathur

Nel

Rais

(2013). Maturation of the adolescent brain. Neuropsychiatr Dis Treat, 9:449–461. https://doi.org/10.2147/NDT.S39776 (Accessed 03/02/2022).

Baker

(2013). Conduct disorders in children and adolescents. Paediatrics and Child Health, 23(1), 9–24. https://doi.org/10.1016/j.paed.2012.09.007 [Accessed 30/03/2022].

Barker

L. F.

Ireland

J. L.

Chu

Ireland

C. A.

(2016). Sleep and its association with aggression among prisoners: Quantity or quality? International Journal of Law and Psychiatry, 47:115–121. https://doi.org/10.1016/j.ijlp.2016.02.014 (Accessed 17/12/2021).

Billingham

Irwin-Rogers

Muthoo

(2020). Youth Violence Commission Final Report. (publisher not known).

Clinkinbeard

S. S.

Simi

Evans

M. K.

Anderson

A. L.

(2011). Sleep and delinquency: Does the amount of sleep matter? J Youth Adolesc, 40(7), 916–930. https://doi.org/10.1007/s10964-010-9594-6 (Accessed 27 11 2021).

Connolly

E. J.

Jackson

D. B.

Semenza

D. C.

(2021). Quality over quantity? Using sibling comparisons to examine relations between sleep quality, sleep duration, and delinquency. Soc Sci Med, 280:114053. https://doi.org/10.1016/j.socscimed.2021.114053 (Accessed 06/01/2022).

Crowley

S. J.

Acebo

Carskadon

M. A.

(2007). Sleep, circadian rhythms, and delayed phase in adolescence. Sleep Med, 8(6), 602–612. https://doi.org/10.1016/j.sleep.2006.12.002 (Accessed 16 12 2021).

Dai

X.-J.

Liu

C.-L.

Zhou

R.-L.

(2015). Long-term total sleep deprivation decreases the default spontaneous activity and connectivity pattern in healthy male subjects: A resting-state fMRI study. Neuropsychiatr Dis Treat, 11:761–772. https://doi.org/10.2147/NDT.S78335 (Accessed 20/03/2022).

10.

de Bruin

E. J.

Bögels

S. M.

Oort

F. J.

Meijer

A. M.

(2018). Improvements of adolescent psychopathology after insomnia treatment: Results from a randomized controlled trial over 1 year. J Child Psychol Psychiatry, 59(5), 509–522. https://doi.org/10.1111/jcpp.12834 (Accessed 22 02 2022).

11.

El-Sheikh

Saini

E. K.

Gillis

B. T.

Kelly

R. J.

(2019). Interactions between sleep duration and quality as predictors of adolescents’ adjustment. Sleep Health, 5(2), 180–186. https://doi.org/10.1016/j.sleh.2018.11.004 (Accessed 25/11/2021).

12.

Fitzsimons

Agalioti-Sgompou

Calderwood

(2020). Millennium cohort study sixth survey 2015-2016 user guide (2nd ed.). University College London: Centre for Longitudinal Studies: Institute of Education.

13.

Freitag

Ireland

J. L.

Niesten

I. J. M.

(2017). Exploring the relationship between sleep quality, emotional well-being and aggression levels in a European sample. J Aggress Confl Peace Res, 9(3), 167–177. https://doi.org/10.1108/jacpr-08-2016-0239 (Accessed 15 03 2022).

14.

Friedman

Taraban

Sitnick

Shaw

D. S.

(2021). Early adolescent predictors of violent behavior: Child and contextual risk, and moderation by rejecting parenting. J Early Adolesc, 41(8), 1228–1254. https://doi.org/10.1177/02724316211002268 (Accessed 15 02 2022).

15.

Granvik Saminathen

Plenty

Modin

(2021). The role of academic achievement in the relationship between school ethos and adolescent distress and aggression: A study of ninth grade students in the segregated school landscape of stockholm. J Youth Adolesc, 50(6), 1205–1218. https://doi.org/10.1007/s10964-020-01199-w (Accessed 10/12/2021).

16.

Haynes

P. L.

Bootzin

R. R.

Smith

Cousins

Cameron

Stevens

(2006). Results from an integrative behavioral sleep- treatment pilot program. Sleep, 29(4), 512–520. https://doi.org/10.1093/sleep/29.4.512 (Accessed 20 03 2022).

17.

T. C.

Connolly

C. G.

Henje Blom

(2015). Emotion-dependent functional connectivity of the default mode network in adolescent depression. Biol Psychiatry, 78(9), 635–646. https://doi.org/10.1016/j.biopsych.2014.09.002 (Accessed 29 03 2022).

18.

Ireland

J. L.

Culpin

(2006). The relationship between sleeping problems and aggression, anger, and impulsivity in a population of juvenile and young offenders. J Adolesc Health, 38(6), 649–655. https://doi.org/10.1016/j.jadohealth.2005.05.027 (Accessed 03/01/2021).

19.

Karriker-Jaffe

K. J.

Foshee

V. A.

Ennett

S. T.

Suchindran

(2008). The development of aggression during adolescence: Sex differences in trajectories of physical and social aggression among youth in rural areas. J Abnorm Child Psychol, 36(8), 1227–1236. https://doi.org/10.1007/s10802-008-9245-5 (Accessed 16 12 2021).

20.

Kohyama

(2021). Which is more important for health: Sleep quantity or sleep quality? Children, 8(7), 542. https://doi.org/10.3390/children8070542 (Accessed 16/02/2022).

21.

Kramer

A. A.

Zimmerman

J. E.

(2007). Assessing the calibration of mortality benchmarks in critical care: The Hosmer- Lemeshow test revisited. Crit Care Med, 35(9), 2052. https://doi.org/10.1097/01.CCM.0000275267.64078.B0 (Accessed 29 03 2022).

22.

Krizan

Herlache

A. D.

(2016). Sleep disruption and aggression: Implications for violence and its prevention. Psychol Violence, 6(4), 542–552. https://doi.org/10.1037/vio0000018 [Accessed 06/12/2021].

23.

Lansford

J. E.

Skinner

A. T.

Sorbring

(2012). Boys’ and girls’ relational and physical aggression in nine countries: Relational and physical aggression. Aggress Behav, 38(4), 298–308. https://doi.org/10.1002/ab.21433 (Accessed 16 12 2021).

24.

Madrid-Valero

J. J.

Ordoñana

J. R.

Klump

K. L.

Burt

S. A.

(2019). Children sleep and antisocial behavior: Differential association of sleep with aggression and rule-breaking. J Abnorm Child Psychol, 47(5), 791–799. https://doi.org/10.1007/s10802-018-0480-0 (Accessed 18 03 2022).

25.

Meerlo

Sgoifo

Suchecki

(2008). Restricted and disrupted sleep: Effects on autonomic function, neuroendocrine stress systems and stress responsivity. Sleep Med Rev, 12(3), 197–210. https://doi.org/10.1016/j.smrv.2007.07.007 (Accessed 10 03 2022).

26.

Mistlberger

R. E.

Skene

D. J.

(2004). Social influences on mammalian circadian rhythms: Animal and human studies. Biol Rev Camb Philos Soc, 79(3), 533–556. https://doi.org/10.1017/s1464793103006353 (Accessed 02/11/22).

27.

Najman

J. M.

Hayatbakhsh

M. R.

McGee

T. R.

Bor

O’Callaghan

M. J.

Williams

G. M.

(2009). The impact of puberty on aggression/delinquency: Adolescence to young adulthood. Aust N Z J Criminol, 42(3), 369–386. https://doi.org/10.1375/acri.42.3.369

28.

Pikard

Roberts

Groll

(2018). Pediatric referrals for urgent psychiatric consultation: Clinical characteristics, diagnoses and outcome of 4 to 12 year old children. J Can Acad Child Adolesc Psychiatry, 27(4), 245–251. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6254263/ (Accessed 10 12 2021).

29.

Pilcher

J. J.

Ginter

D. R.

Sadowsky

(1997). Sleep quality versus sleep quantity: Relationships between sleep and measures of health, well-being and sleepiness in college students. J Psychosom Res, 42(6), 583–596. https://doi.org/10.1016/s0022-3999(97)00004-4 (Accessed 15 02 2022).

30.

Roman

Walstra

Luiten

P. G. M.

Meerlo

(2005). Too little sleep gradually desensitizes the serotonin 1A receptor system. Sleep, 28(12), 1505–1510. https://doi.org/10.1093/sleep/28.12.1505 (Accessed 11 03 2022).

31.

Stone

L. L.

Janssens

JMAM

Vermulst

A. A.

Van Der Maten

Engels

RCME

Otten

(2015). The strengths and Difficulties questionnaire: Psychometric properties of the parent and teacher version in children aged 4-7. BMC Psychol, 3(1), 4. https://doi.org/10.1186/s40359-015-0061-8 [Accessed 10/12/2021].

32.

Strenziok

Krueger

Heinecke

(2011). Developmental effects of aggressive behavior in male adolescents assessed with structural and functional brain imaging. Soc Cogn Affect Neurosci, 6(1), 2–11. https://doi.org/10.1093/scan/nsp036 (Accessed 24 03 2022).

33.

Sung

Park

Kim

S.-Y.

(2020). Structural alterations in large-scale brain networks and their relationship with sleep disturbances in the adolescent population. Sci Rep, 10(1), 3853. https://doi.org/10.1038/s41598-020-60692-1 (Accessed 22 03 2022).

34.

Suni

(2022a). Teens and sleep: an overview of why teens face unique sleep challenges and tips to help them sleep better. Sleep Foundation. Available from: https://www.sleepfoundation.org/teens-and-sleep (Accessed 05/03/2022).

35.

Suni

(2022b). How to determine poor sleep quality. Sleep Foundation. Available from: https://www.sleepfoundation.org/sleep-hygiene/how-to-determine-poor-quality-sleep (Accessed 10 03 2022).

36.

Tashjian

S. M.

Goldenberg

Monti

M. M.

Galván

(2018). Sleep quality and adolescent default mode network connectivity. Social Cognitive and Affective Neuroscience, 13(3), 290–299. https://doi.org/10.1093/scan/nsy009 (Accessed 20/03/2022).

37.

Terapia Centre. (2020). Scoring strengths and Difficulties questionnaire for age 4-17 or 18+. Terapia Centre. Available from: https://terapia.co.uk/wp-content/uploads/2020/05/SDQ-scoring_Instructions_4-18-years.pdf (Accessed 10 12 2021).

38.

Tremaine

R. B.

Dorrian

Blunden

(2010). Subjective and objective sleep in children and adolescents: Measurement, age, and gender differences: Measurement of child and adolescent sleep. Sleep Biol Rhythms, 8(4), 229–238. https://doi.org/10.1111/j.1479-8425.2010.00452.x (Accessed 04 02 2022).

39.

van Rooij

Hartman

C. A.

Mennes

(2015). Altered neural connectivity during response inhibition in adolescents with attention-deficit/hyperactivity disorder and their unaffected siblings. NeuroImage Clin, 7:325–335. https://doi.org/10.1016/j.nicl.2015.01.004 (Accessed 29/03/2022).

40.

Van Veen

M. M.

Lancel

Beijer

Remmelzwaal

Rutters

(2021). The association of sleep quality and aggression: A systematic review and meta-analysis of observational studies. Sleep Med Rev, 59(101500), 101500. https://doi.org/10.1016/j.smrv.2021.101500 (Accessed 12 12 2021).

41.

Vermeulen

M. C. M.

van der Heijden

K. B.

Kocevska

(2021). Associations of sleep with psychological problems and well-being in adolescence: Causality or common genetic predispositions? Journal of Child Psychology and Psychiatry, 62(1), 28–39. https://doi.org/10.1111/jcpp.13238 (Accessed 10/01/2022).

42.

Woerner

Fleitlich-Bilyk

Martinussen

(2004). The strengths and Difficulties questionnaire overseas: Evaluations and applications of the SDQ beyond europe. Eur Child Adolesc Psychiatry, 13(S2), 47–54. https://doi.org/10.1007/s00787-004-2008-0 (Accessed 02/12/2021).

43.

World Health Organisation (author not known) . (2020). Youth Violence. Who. Int. Available from: https://www.who.int/news-room/fact-sheets/detail/youth-violence? (Accessed 12 04 2022).

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.30 MB