External Publications Using GUI Data

Background
Physical activity plays an important role in optimising physical and mental health during childhood, adolescence, and throughout adult life. This study aims to identify individual, family and environmental factors that determine physical activity levels in a population sample of children in Ireland.

Methods
Cross-sectional analysis of the first wave (2008) of the nationally representative Growing Up in Ireland study. A two-stage clustered sampling method was used where national schools served as the primary sampling unit (response rate: 82%) and age eligible children from participating schools were the secondary units (response rate: 57%). Parent reported child physical activity levels and potential covariates (parent and child reported) include favourite hobby, total screen time, sports participation and child body mass index (measured by trained researcher). Univariate and multivariate multinomial logistic regression (forward block entry) examined the association between individual, family and environmental level factors and physical activity levels.

Results
The children (N = 8,568) were classified as achieving low (25%), moderate (20%) or high (55%) physical activity levels. In the fully adjusted model, male gender (OR 1.64 [95% CI: 1.34-2.01]), having an active favourite hobby (OR 1.65 [95% CI: 1.31-2.08]) and membership of sports or fitness team (OR 1.90 [95% CI: 1.48-2.45]) were significantly associated with being in the high physical activity group. Exceeding two hours total screen time (OR 0.66 [95% CI: 0.52-0.85]), being overweight (OR 0.41 [95%CI: 0.27-0.61]; or obese (OR 0.68 [95%CI: 0.54-0.86]) were significantly associated with decreased odds of being in the high physical activity group.

Conclusions
Individual level factors appear to predict PA levels when considered in the multiple domains. Future research should aim to use more robust objective measures to explore the usefulness of the interconnect that exists across these domains. In particular how the family and environmental settings could be useful facilitators for consistent individual level factors such as sports participation.

The microbiome-gut-brain axis plays a role in anxiety, the stress response and social development, and is of growing interest in neuropsychiatric conditions. The gut microbiota shows compositional alterations in a variety of psychiatric disorders including depression, generalised anxiety disorder (GAD), autism spectrum disorder (ASD) and schizophrenia but studies investigating the gut microbiome in social anxiety disorder (SAD) are very limited. Using whole-genome shotgun analysis of 49 faecal samples (31 cases and 18 sex- and age-matched controls), we analysed compositional and functional differences in the gut microbiome of patients with SAD in comparison to healthy controls. Overall microbiota composition, as measured by beta-diversity, was found to be different between the SAD and control groups and several taxonomic differences were seen at a genus- and species-level. The relative abundance of the genera Anaeromassillibacillus and Gordonibacter were elevated in SAD, while Parasuterella was enriched in healthy controls. At a species-level, Anaeromassilibacillus sp An250 was found to be more abundant in SAD patients while Parasutterella excrementihominis was higher in controls. No differences were seen in alpha diversity. In relation to functional differences, the gut metabolic module ‘aspartate degradation I’ was elevated in SAD patients. In conclusion, the gut microbiome of patients with SAD differs in composition and function to that of healthy controls. Larger, longitudinal studies are warranted to validate these preliminary results and explore the clinical implications of these microbiome changes.

Specific pregnancy complications, socioeconomic position and sex have all been independently associated with child mental health outcomes, but their combined effects remain unclear. We examined whether total number of complications experienced in the pregnancy associated with mental health at 5 and 9-years, and whether this varied by sex or adverse social circumstances. Pregnancy complications were self-reported at 9-months post-natally from a list of 16 complications. Parents completed the Strengths and Difficulties Questionnaire (SDQ) when their child was 5 and 9-years. The primary outcome was the SDQ-total and scoring in the clinical range (> 16) was a secondary outcome. We applied generalized linear mixed models to a large nationally representative Irish cohort (GUI; n = 11,134). Analyses were adjusted for sex, adverse social circumstances (at 9-months), and gestational smoking. We included an interaction term between pregnancy complications and each variable respectively in separate models to examine if associations varied by sex or adverse circumstances.

After controlling for covariates, total complications associated with mental health at 5 and 9-years. Each additional pregnancy complication conferred a 10% higher total-SDQ score (exponentiated co-efficient 1.10 [95%CI 1.06–1.14], 1.20 [1.15–1.26], 1.20 [1.12–1.29] and 1.34 [1.21–1.48] for 1, 2, 3 and 4 + complications respectively). For the dichotomised outcome, generally increasing odds for clinical levels of mental health difficulties were observed (OR 1complication = 1.89, 95%CI [1.37–2.59]; OR 2complications = 2.31, 95%CI [1.53–3.50]; OR 3complications = 1.77, 95%CI [0.89–3.52]; OR 4 + complications = 6.88, 95%CI [3.29–14.40]). Females had significantly lower odds of exhibiting clinically significant mental health difficulties than males (OR = 0.43, 95%CI[0.32–0.57]).

There was no evidence that the association between pregnancy complications and child’s mental health varied by sex or social circumstances at 5 or 9-years. Males exposed to numerous pregnancy complications in the context of adverse social circumstances had the highest predicted probability of having mental health difficulties in middle childhood.

The aim of this paper is to quantify the effect of economic, parental and lifestyle factors on the psychological health of children at significant points in recent economic history in Ireland. The paper uses data from the Growing Up in Ireland study and employs a dynamic random effects ordered logistic regression model to test the magnitude of these effects. The results indicate that proxy income variables, such as ability to make ends meet and homeownership, are relevant in predicting child psychological health outcomes. Equally important are intergenerational health associations between the mother and the psychological health of children.

The incidence of adiposity in the early years of life has outgrown the prevalence in older children and adolescents globally; however, the factors influencing weight gain are predominantly studied in adults. This study examines the relationship between changing economic conditions during the Irish recession and child weight in a nationally representative sample of 10,011 Irish children studied at 9 months, 3 years and 5 years old. This study is the first to use longitudinal anthropometric measurements to estimate the impact of direct parental unemployment on children’s weight. Fixed effect logistic regression is used to examine the effects of parental unemployment on weight using the Growing up in Ireland infant cohort from 2008 to 2013. Weight and length/height measured by trained interviewers and child growth charts are used to quantify children’s weight status according to BMI-for-age and weight-for-age measures. For BMI-for-age, the probability of a child being classified as overweight/obese is 8 percentage points higher if either parent experiences unemployment. A sensitivity analysis of weight-for-age indicates that the probability is of similar magnitude across several alternative growth charts and definitions of adiposity (the WHO standard, British Growth Reference, and Centers for Disease Control). The analysis is repeated, cross-sectionally, for physical activity and diet to clarify mechanisms of effect. The probability of a child consuming healthy food and physical activity with an implied cost is lower if either parent becomes unemployed. A focus on overweight/obesity in the early years is of crucial importance as if current trends are not addressed, a generation of children may grow up with a higher level of chronic disease.

Background
Early and risky adolescent alcohol use have each been associated with adult alcohol consumption. However, it remains unclear whether these behaviours independently predict later-life substance use when considered jointly, and research examining links with substances other than alcohol is limited. This study addresses these gaps by examining longitudinal associations between age at first alcohol and risky adolescent alcohol use, with alcohol, tobacco, cannabis and other drug use in early adulthood, and aims to identify critical periods for public health interventions.

Methods
Growing Up in Ireland is a nationally representative cohort (recruited aged 9 [Wave 1], born 1997–1998). Survey-weighted logistic regression examined whether age at first alcoholic drink and risky alcohol use at age 17 (Alcohol Use Disorders Identification Test scores) independently predict high-risk alcohol (AUDIT > 15), tobacco, cannabis and other drug use at age 20. Models were adjusted for age, sex, academic ability, personality, psychological factors, socioeconomic status, familial, peer and neighbourhood substance use.

Results
The study included 4554 participants (49.8% female). Early alcohol use was common, with 27% reporting use aged 14 or younger. By age 20, 14% reported high-risk alcohol, 38% tobacco, 24% cannabis and 28% other drug use. Older age at first alcohol was associated with dose-response reductions in the odds of high-risk alcohol, tobacco, cannabis and other drug use at age 20, relative to those initiating alcohol at 14 or younger. Adolescents with high-risk alcohol use had double the odds of tobacco (adjusted odds ratio (aOR) 2.1, 95% confidence intervals (CI) 1.3–3.30) and other drug use (aOR 2.5, 95% CI 1.6–4.1) and an 11-fold increase in the odds of continued high-risk alcohol use (aOR 11.5, 95% CI 7.0–18.6) at age 20, relative to adolescents with low-risk alcohol use.

Conclusions
Age at first alcohol and risky adolescent alcohol use independently predict substance use in early adulthood when considered jointly in extensively adjusted models. These findings highlight the continued urgent need for public health interventions that address factors associated with early alcohol use and support adolescents who use alcohol in a high-risk manner given their elevated risk of progression to more serious substance use as adults.

Purpose
Despite growing concerns about trends in cocaine use, there is a shortage of longitudinal research that prospectively examines risk and protective factors associated with cocaine initiation and use in general youth populations. This study addresses this gap.

Methods
Growing Up in Ireland is a nationally representative cohort. Individual, family, and socio-environmental exposures associated with incident past-year cocaine use at ages 17 (N = 5965) and 20 (n = 4549) were assessed with survey-weighted logistic regression using generalised estimating equations. Prevalent past-year cocaine use at 20 (N = 4679) was analysed using generalised estimating equations complemented by gradient-boosted decision trees and Shapley explanations.

Results
221 (3.7%) self-reported cocaine use at 17 and 1072 (22.9%) at 20. Alcohol use at 14 or younger was associated with eight times the odds of cocaine use at 17 (aOR 8.0, 95% CI 1.7–37.3) and 19 times at 20 (aOR 19.2, 95% CI 8.6–43.2). Peer cannabis use was associated with 7 times the odds of cocaine use at 17 (aOR 7.3, 95% CI 2.9–18.3) and double at 20 (aOR 2.4, 95% CI 1.8–3.2). Growing up in a neighbourhood where substance use was common doubled the odds of cocaine use at 17 (aOR 2.4, 95% CI 1.3–4.4). Shapley explanations revealed individual-specific positive or negative impacts of exposures.

Discussion
Cocaine use among 20-year-olds in Ireland is higher than reported internationally, and increases sharply between the ages of 17 and 20, suggesting a need for interventions targeting this age group. However, associations with early adolescent factors suggest that early interventions may also be important.

Objective
This study aims to examine the associations between prenatal exposure to maternal smoking, birth weight and persistent offspring psychiatric symptoms. Additionally, we aim to examine whether the relationship between prenatal maternal smoking and persistent offspring psychiatric symptoms is mediated by offspring birth weight.

Methods
This study used the Growing Up in Ireland (GUI) longitudinal cohort. The GUI is a nationally representative longitudinal study of children which consisted of three data collection waves, at ages 9, 13, and 17 years. Logistic regression analysis was used to examine associations between prenatal tobacco exposure, and offspring psychiatric symptoms. Linear regression was used to examine associations between prenatal tobacco exposure and offspring birth weight. We conducted a mediation analysis examining potential etiological pathways linking maternal smoking during pregnancy, offspring birth weight, and later offspring psychiatric symptoms. All analyses were adjusted for confounders including household income, maternal level of education, and family psychiatric history. Additionally, examination of birth weight and subsequent psychiatric symptoms also was controlled for prematurity.

Results
We found that the association between prenatal tobacco exposure and later psychiatric symptoms is mediated by birth weight.

Conclusions
This work provides further evidence that maternal smoking during pregnancy is an important modifiable lifestyle factor that has an impact not just on the physical health of offspring but also their mental wellbeing. Supporting women with structured smoking cessation programs at the earliest stages of pregnancy should be a public health priority.

Objectives
To assess the impact of digital media usage on psychiatric symptoms in an adolescent population utilizing a longitudinal cohort design.

Methods
Using two waves of the GUI child cohort, age 13 (N = 7527) and age 17/18 (N = 6126), we used latent class analysis (LCA) to create latent groups centred around self-reported time spent online, and the self-reported behaviours children engaged with online. At both waves, the 4 class latent model suited best. We used the different symptoms scales of the Strengths and Difficulties Questionnaires (SDQ), self-reported, at age 17 as our outcome variable. Using linear regression analysis, we then examined the associations between our latent class model and psychiatric symptoms, using moderate usage as our reference group, with adjustments being made for baseline psychiatric symptoms, maternal education and pre-diagnosed mental disorder.

Results
For females, placement in the high usage group at 13 was associated with increased internalizing symptoms, whereas placement in the high usage group at 17 was associated with an increase in all symptoms. For males, placement in the high usage group at 17 was associated with increased emotional symptoms, and placement in the “low usage & behaviour engagement”group, a group showing low reported time online and low engagement in our measured online behaviours at 17, was associated with an increase in all symptoms. Finally for both sexes, placement in the “moderate usage, entertainment only” group at age 13, (a group reporting no school-based online engagement), was associated with increases in all symptoms except emotional symptoms.

Conclusion
High digital media usage is associated with increased psychiatric symptoms in both males and females, with moderate usage associated with positive effects on symptoms compared to both our high usage, and low usage groups.

Keywords
Digital engagement, Psychiatric symptoms, Internet usage, Latent class analysis

Aims
This study aims to replicate Przybylski and Weinstein (Psychol Sci 28(2):204–215, 2017), using a large population cohort to examine the validity of the proposed Goldilocks Hypothesis, which states that moderate digital media engagement may be beneficial and that both high and low usage may have a negative relationship with mental wellbeing.

Methods
Using the GUI98 cohort, we used separate weekday and weekend time-based categorical variables indicating time spent online, playing video games, watching TV/films as well as a frequency variable indicating multiscreening, and their associations with SDQ internalizing and externalizing symptoms using linear and quadratic regression parameters. We followed procedures for confounder adjustments outlined in Przybylski and Weinstein (Psychol Sci 28(2):204–215, 2017).

Results
As hypothesized by the Goldilocks Hypothesis, time spent online watching TV/films at the weekend and multiscreening all had curvilinear relationships with internalizing and externalizing symptoms with significantly higher symptoms for no time as well as for higher exposures. internalizing and externalizing symptoms increased with time spent playing video games.

Conclusions
This brief report supports the Goldilocks Hypothesis, that suggests that moderate use of digital technology is not intrinsically harmful and may instead be beneficial, even necessary in a world becoming ever more increasingly reliant on digital media (Przybylski and Weinstein in Psychol Sci 28(2):204–215, 2017).

Previous research has suggested that the children of teenage parents are at a high risk of developing both internalising and externalising behaviour disorders. The current study aimed to explore pathways through which children of teenage mothers show more externalising and internalising psychopathology than their peers whose parents were older. The present study used data from the first wave of the ‘Growing Up in Ireland’ 9-year-old cohort and employed structural equation modelling to assess the explanatory value of a model informed by previous research findings. Goodness-of-fit indices indicated that many aspects of the familial environment of Irish teenage families did indeed place children at a substantial risk of problematic developmental outcomes. In particular, the results implicated the influential power of economic stability and familial conflict on the well-being of teenage families and related child outcomes, results similar to those reported in international studies.

Keywords: teenage mothers in Ireland, economic stability, externalising problems, internalising problems

Background
Parental incarceration (PI) is associated with adverse developmental outcomes for children affected. However, research in this area often reports conflicting results with few studies following children across time in non-U.S. populations. Additionally, more research is called for using multi-informant perspectives rather than relying on adult reports of child outcomes alone.

Methods
This study used data from the first two waves of a nationally representative cohort study of 8,568 children aged 9 years and followed up at age 13 living in the Republic of Ireland (2007–2012). Propensity score matching was used to match children who had experienced PI by the age of nine to children who had not experienced PI by sociodemographics and experience of other stressful events. Mental health, as measured by self-concept (Piers-Harris II) and externalizing and internalizing difficulties (strength and difficulties questionnaire), was compared across both groups.

Results
Fifty of the 8,568 children (weighted percentage 0.9%) reported experiencing PI by the age of nine. These children came from more socially disadvantaged homes and were more likely to have experienced other potentially stressful life events. In comparison to a matched sample of children not affected by PI, children affected by PI reported higher levels of anxiety at age nine. Longitudinal analysis indicated these children affected by PI also reported lower levels of happiness at age 13 with higher levels of emotional difficulties reported by their primary caregiver.

Conclusions
Children of incarcerated parents face a greater array of life challenges. PI had an association with child-reported levels of anxiety at age nine. PI also had a medium-term association on caregiver assessments of emotional difficulties of children affected as well child-reported levels of happiness over time.

Background
There are no early, accurate, scalable methods for identifying infants at high risk of poor cognitive outcomes in childhood. We aim to develop an explainable predictive model, using machine learning and population-based cohort data, for this purpose.

Methods
Data were from 8858 participants in the Growing Up in Ireland cohort, a nationally representative study of infants and their primary caregivers (PCGs). Maternal, infant, and socioeconomic characteristics were collected at 9-months and cognitive ability measured at age 5 years. Data preprocessing, synthetic minority oversampling, and feature selection were performed prior to training a variety of machine learning models using ten-fold cross validated grid search to tune hyperparameters. Final models were tested on an unseen test set.

Results
A random forest (RF) model containing 15 participant-reported features in the first year of infant life, achieved an area under the receiver operating characteristic curve (AUROC) of 0.77 for predicting low cognitive ability at age 5. This model could detect 72% of infants with low cognitive ability, with a specificity of 66%.

Conclusions
Model performance would need to be improved before consideration as a population-level screening tool. However, this is a first step towards early, individual, risk stratification to allow targeted childhood screening.

Impact
• This study is among the first to investigate whether machine learning methods can be used at a population-level to predict which infants are at high risk of low cognitive ability in childhood.
• A random forest model using 15 features which could be easily collected in the perinatal period achieved an AUROC of 0.77 for predicting low cognitive ability.
• Improved predictive performance would be required to implement this model at a population level but this may be a first step towards early, individual, risk stratification.

Background
At age 5, while only embarking on their educational journey, substantial differences in children’s cognitive ability will already exist. The aim of this study was to examine the causal association between below average cognitive ability at age 5 years and child-reported experience of school and self-concept, and teacher-reported class engagement and emotional-behavioural function at age 9 years.

Methods
This longitudinal cohort study used data from 7,392 children in the Growing Up in Ireland Infant Cohort, who had completed the Picture Similarities and Naming Vocabulary subtests of the British Abilities Scales at age 5. Principal components analysis was used to produce a composite general cognitive ability score for each child. Children with a general cognitive ability score more than 1 standard deviation (SD) below the mean at age 5 were categorised as ‘Below Average Cognitive Ability’ (BACA), and those scoring above this as ‘Typical Cognitive Development’ (TCD). The outcomes of interest, measured at age 9, were child-reported experience of school, child’s self-concept, teacher-reported class engagement, and teacher-reported emotional behavioural function. Binary and multinomial logistic regression models were used to examine the association between BACA and these outcomes.

Results
Compared to those with TCD, those with BACA had significantly higher odds of never liking school [Adjusted odds ratio (AOR) 1.82, 95% CI 1.37–2.43, p < 0.001], of being picked on (AOR 1.27, 95% CI 1.09–1.48) and of picking on others (AOR 1.53, 95% CI 1.27–1.84). They had significantly higher odds of experiencing low self-concept (AOR 1.20, 95% CI 1.02–1.42) and emotional-behavioural difficulties (AOR 1.34, 95% CI 1.10–1.63, p = 0.003). Compared to those with TCD, children with BACA had significantly higher odds of hardly ever or never being interested, motivated and excited to learn (AOR 2.29, 95% CI 1.70–3.10).

Conclusion
Children with BACA at school-entry had significantly higher odds of reporting a negative school experience and low self-concept at age 9. They had significantly higher odds of having teacher-reported poor class engagement and problematic emotional-behavioural function at age 9. The findings of this study suggest BACA has a causal role in these adverse outcomes. Early childhood policy and intervention design should be cognisant of the important role of cognitive ability in school and childhood outcomes.

Children with below average cognitive function represent a substantial yet under-researched population for whom academic and social demands, which increase in complexity year by year, pose significant challenge. Effects on emotional-behavioural development (EBD) are not well understood. The aim of this study was to compare trajectories of EBD for children with and without below average cognitive function. The underlying hypothesis was that trajectories of EBD would differ between groups, with divergence occurring as children are subject to increasingly complex cognitive demands.

Participants consist of 7,000 children and caregivers who completed the Growing Up in Ireland survey at age 3,5 and 9 years. Cognitive function was measured at age 3 using the Picture Similarities Scale. A t-score 1-2 standard deviations below the mean was categorised as below average cognitive function (n=767), and scores above this categorised as average cognitive function (n=6418). EBD was measured using the Strengths and Difficulties Questionnaire (SDQ) at age 3,5 and 9. Repeated measures ANOVA was used to examine the difference between the overall group means across the repeated measures of SDQ, the SDQ change over time, and the interaction between cognitive group and change over time. Further analysis of trajectories was performed using latent growth curve analysis.

Compared to those with average cognitive function, a significantly higher proportion of children with below average function were male (61.8% v 38.2%, χ2 (1, N=7134) = 42.07 p<0.001), born to a single parent family (22.4% v 14.7%, χ2 (1, N=7134) = 23.15 p<0.001), and had a parent who smoked (35.6% v 25.8%, χ2 (1, N=7134) = 33.23 p<0.001). Children with below average cognitive function had significantly higher mean total SDQ scores at all ages. Repeated measures ANOVA demonstrated a significant group-by-time interaction effect (F(2,7182)=4.649, p=0.010). The mean difference (MD) in SDQ between cognitive groups increased over time (MD Age 3:0.87, 95% confidence interval (CI) 0.53-1.21, Age 9 MD:1.49,95% CI 1.08-1.91). For those with average or above cognitive function the overall SDQ decreased between age 3-9 (MD: -0.49 95% CI -0.65—0.33). This decrease was not seen for those with below average function, who had a non-statistically significant increase in SDQ between age 3-9.

Children with below average cognitive function experience higher and worsening mean SDQ scores throughout childhood. A scalable method of early identification of children at risk for below average cognitive function should be a research priority for public health, enabling early intervention for cognitive and adaptive outcomes.

The first 1000 days of life are a period of unique sensitivity and plasticity during which critical cognitive abilities are formed. Routine developmental screening tools aim to identify infants who would benefit from early intervention. While these tools have been validated for detecting children with more severe neurodevelopmental disorders, their ability to identify the larger proportion with below average cognitive function has not been sufficiently explored. The aim of this study was to examine the validity of the Ages and Stages Questionnaire (ASQ), for identifying children with later below average cognitive function.

The study population (n=8260) is formed from two national cohort studies, the Growing Up in Ireland (GUI) Infant cohort (n=7,444) and the Cork BASELINE cohort (n=816). The ASQ was completed at 8 months and 24-27 months respectively. Cognitive assessments were performed at age 5. Those scoring <1 standard deviation (SD) below the mean were categorised as below average cognitive function. Applying the currently used onward referral criterion (one fail in any domain) the sensitivity, specificity, positive and negative predictive values of the 8- and 24-27- month ASQ for detecting children with later below average cognitive function were calculated.

In the GUI cohort n=905 participants (12.5%) had scores <1SD below the mean on the Picture Similarities Scale. In the BASELINE cohort n=101 participants (13.4%) had an IQ <1SD below the cohort mean. Applying the currently used onward referral criterion (failing in any one domain in the ASQ), the sensitivity of the 8-month ASQ for detecting children scoring <1SD below the mean on the Picture Similarities Scale at age 5 was 16.4% (95% CI 14.0-19.0). The specificity was 92.0% (95% CI 91.3-92.6), with a positive predictive value (PPV) of 22.6% (95% CI 19.5-26.0) and a negative predictive value (NPV) of 88.5% (95% CI 87.7%-89.2%).

In the BASELINE cohort n=468 participants completed the 24-month ASQ and n=316 the 27-month ASQ. Applying the same onward referral criterion to the 24- and 27- month ASQ combined, the sensitivity for detecting those with an IQ <1SD below the cohort mean was 20.8% (95% CI 13.6-30.2) and the specificity was 91.1% (95% CI 88.6-93.2).

The ASQ has a low sensitivity for identifying children with below average cognitive function at age 5. The findings of this study suggest that if we are to intervene early in the developmental trajectory for children with below average cognitive function alternative methods of identifying high risk infants are needed.

Background
There is growing concern around youth mental health. A population health approach to improve mental health must address, among other issues, economic insecurity, access to housing and education, harm reduction from substance use. As a universal public health intervention, increasing physical activity at a population level may have an important role in our approach. The aim of this study was to examine the longitudinal association between physical activity patterns between childhood and early adolescence and emotional-behavioural difficulties in later adolescence.

Methods
This study was based on data from the ′98 Child cohort of the Growing Up in Ireland Study. Participants were categorized according to physical activity levels at ages 9 and 13. Emotional-behavioural difficulties at age 17 were measured using the parent-reported Strengths and Difficulties Questionnaire. Logistic regression was used to examine the association between physical activity and emotional-behavioural outcomes.

Results
Among 4618 participants included in the regression model, those categorized as Inactive (n=1607) or Reducer (n=1662) were more than twice as likely to have emotional-behavioural difficulties at age 17 compared with those who were Active [adjusted odds ratio (AOR) 2.1, 95% CI 1.46–3.01, P<0.001; AOR 1.93, 95% CI 1.34–2.76, P<0.001, respectively]. Among those with emotional-behavioural difficulties at baseline (n=525), those categorized as Active had 2.3-fold reduced odds for emotional-behavioural problems at age 17 compared with those who were Inactive (AOR 0.43, 95% CI 0.23–0.78, P=0.006).

Conclusions
Increasing physical activity among adolescents is a safe and sustainable public health intervention associated with improved mental health.