External Publications Using GUI Data

Growing Up in Ireland (GUI) is a two-cohort, longitudinal study of children and young people. The study aims to describe the health and development of Irish children across a range of topics; these include physical and mental health, family socio-demographic status, education, and the child’s behaviour, attitudes and key relationships. The study has been collecting data since 2007, beginning with a child cohort at nine years old (n = 8,568) and then an infant cohort at nine months old (n = 11,134). These data provide researchers and policy makers with a unique analytical tool to explore the well-being of children in Ireland. This paper provides an overview of all the stages involved in the development of the study, from its inception, to the establishment of the study’s aims, objectives and design, the ongoing data collection and panel maintenance, and the many uses of GUI data today.

To describe response rates and characteristics associated with response to the Pregnancy Risk Assessment Monitoring System study in Ireland (PRAMS Ireland). Using hospital discharge records of live births at a large, urban, obstetric hospital, a sampling frame of approximately 2,400 mother-infant pairs were used to alternately sample 1,200 women. Mothers’ information including name, address, parity, age and infant characteristics such as sex and gestational age at delivery were extracted from records. Modes of contact included an invitation letter with option to opt out of the study, three mail surveys, a reminder letter and text message reminder for remaining non-respondents. Sixty-one per cent of women responded to the PRAMS Ireland survey over a 133 day response period. Women aged <30, single women, multiparous women and women with a preterm delivery were less likely to respond. Women participating in PRAMS Ireland were similar to the national birth profile in 2011 which had a mean age of 32, were 40 % primiparous, 33 % single or never married and had a 28 % caesarean section rate. Survey and protocol changes are required to increase response rates above recommended Centers for Disease Control and Prevention (CDC) thresholds of 65 % within the recommended 90 day data collection cycle. Additional efforts such as stratification and over-sampling are required to increase representativeness among hard to reach groups such as younger, single and multiparous women before expanding the project to an ongoing, national surveillance system in Ireland.

The present study examines whether breastfeeding is associated with neuro-developmental advantages at 9 months of age on a standardised measure of infant development in a large cohort study of Irish children. It is hypothesised that if breast-milk confers an independent benefit, infants who were never breastfed will have reached fewer developmental milestones than those who were partially or exclusively breastfed, after controlling for putative confounding variables. Families with infants aged 9-months were recruited as part of a nationally representative sample for the birth cohort of the Growing Up in Ireland study (n = 11,134). Information was collected from mothers on breastfeeding practices, socio-demographic characteristics and developmental progress during a household interview. Parent-report items on development covered communication, gross motor, fine motor, problem solving and personal-social skills. Analysis of pass/fail status in each developmental domain using binary logistic regression showed a positive effect of any breastfeeding on gross motor, fine motor, problem solving and personal-social skills (but not communication) and these remained after adjustment for a range of confounding variables. There was, however, little evidence of a dose–response effect or advantage of exclusive over partial breastfeeding. A clear advantage of breastfeeding on infant development was demonstrated. However, the lack of a dose–response association on pass rates suggests that the breastfeeding effect may be confounded by other unobserved factors or that there is a critical threshold during which time the effect of breast milk may be particularly salient for bolstering brain development.

Objectives
To examine the associations between preeclampsia and longitudinal child developmental and behavioural outcomes using data from a nationally representative study of children living in Ireland.

Methods
We used maternal-reported data from the Growing Up in Ireland longitudinal study of children. Data on preeclampsia and preeclampsia + small for gestational age (SGA) were collected when children were 9-months old. Data on child development and behavioural outcomes were collected at 9-months using the Ages and Stages Questionnaire (ASQ), and at 3 years, 5 years and 7–8 years using the Strengths and Difficulties Questionnaire (SDQ). Multivariate logistic regression analysis was used to examine the association between preeclampsia exposure and failure of ASQ domains, and abnormal SDQ domains. Linear spline multilevel models were used to examine the association between preeclampsia and preeclampsia + SGA and repeated measures of SDQ. All models controlled for several perinatal and sociodemographic factors.

Results
A total of 10,692 children were included in the study at baseline, representing a weighted total of 70,791. Multivariate logistic regression suggested that preeclampsia was not associated with failing any ASQ domain. Preeclampsia was associated with abnormal SDQ cut-off of emotional (≥ 5) and hyperactivity (≥ 7) domains at age 5 years only. In the linear spline model, mean SDQ score was higher at each time point in exposed groups.

Conclusions for Practice
While we did not find strong evidence of associations between preeclampsia and child developmental and behavioural outcomes overall, some associations between preeclampsia-exposure and subtle behavioural issues did persist. Further research is needed to replicate these findings, and determine the clinical significance of changes in SDQ scores.

The aim of this study was to identify, using cluster analysis, novel sleep phenotypes in a population based cohort of infants, and to explore the associations between infant sleep profiles and their mothers’ health and well-being. 11,134 mothers of 9-month old infants were interviewed as part of the Growing Up in Ireland National Longitudinal study and reported on their health and infant’s sleep patterns. 16 infant sleep variables were recorded together with measures of parental stress, depression, health and well-being. Multiple iterations of a two-step hierarchical cluster analysis were carried out to identify the optimum number of clusters and the subset of parental-reported sleep variables required to identify distinct sleep profiles. Four distinct sleep profiles were identified based on the following variables; (1) infant sleep duration at night, (2) parental sleep duration, (3) does baby wake during night (yes, no)? (4) Usual sleep location for most of the night and, (5) parental reporting of problem infant sleep patterns. This identified two less favorable profiles with both infants and mothers sleeping less and where mothers are more likely to report their infants’ sleep patterns as problematic. Mothers of infants belonging to these sleep profiles were more likely to have higher levels of stress, depressive symptoms, and poorer self-reported health than other sleep profiles. Breastfeeding was associated with both groups and rates were highest in a group of infants that were more likely to co-sleep with their parents and have diverse ethnic backgrounds. This study demonstrates, for the first time, two infant sleep profiles with distinct phenotypical frameworks that are significantly associated with maternal stress, depression, and poorer self-report of health.

Background
Poverty is associated with poor outcomes, yet exposure to distinct poverty trajectories in early childhood is not well understood.

Objective
To understand the prevalence of different trajectories of household poverty and their association with mid-childhood and primary-parent(mother) indicators of physical health and psychopathology in Ireland.

Methods
We used a nationally representative, prospective cohort (Growing-up-in-Ireland– Infant-Cohort). Household poverty included lowest third income decile, subjective poverty and material deprivation when children were aged 9-months(m),3,5,9-years(y). We used group based multi-trajectory-cluster-modelling to classify trajectories of poverty. Using multivariate logistic regression, adjusted with separate child and mother confounders, we assessed the association of poverty trajectories from 9m to 9y with child outcomes (overweight, any longstanding illness and psychopathology) at age 9y and the same poverty trajectories over the same 9y period with mother outcomes (overweight, any longstanding illness and depression).

Results
Of 11,134 participants, four trajectories were identified: never-in-poverty (43.1%), material/subjective>monetary-poverty (16.1%), monetary>material-poverty (25.6%), and persistent-poverty (15.2%). Children in persistent-poverty compared with never in poverty experienced higher odds of being overweight at 9y (adjusted odds ratio aOR 1.70,95% CI 1.34, 2.16), having a longstanding illness (aOR 1.51,95% CI 1.20, 1.91) and psychopathology (aOR 2.06,95% CI 1.42, 2.99). The outcomes for primary parent (99.7% were mothers) had higher odds of being overweight (aOR 1.49,95% CI 1.16, 1.92), having a longstanding illness (aOR 2.13,95% CI 1.63, 2.79), and depression (aOR 3.54,95% CI 2.54, 4.94).

Conclusions
Any poverty trajectory was associated with poorer psychopathology and physical wellbeing in late childhood for children and their mothers in Ireland.

What is already known on this subject?
Poverty is understood as having both objective (e.g. low household income) and subjective dimensions (e.g., perception of being poor relative to others, poverty related anxiety). Evidence highlights the association between child poverty exposure and child and mother wellbeing outcomes.

What this study adds?
Our study examined predicted trajectories of multi-dimensional poverty and child and mother outcomes using a large national cohort. We identified that any poverty trajectory in childhood had a higher odds of being associated with poor well-being outcomes.

How this study might affect research, practice or policy?
This study highlights the need for strategies to ameliorate any poverty exposure as all poverty trajectories resulted in poorer wellbeing outcomes. Strategies should be considered to ameliorate any poverty exposure in early childhood (e.g., school based resourcing).

The increasing emphasis on inclusive education internationally has broadened the definition of special educational needs (SEN) and greatly affected national prevalence estimates. In line with these international trends, in Ireland the EPSEN Act (2004) defines SEN as any “restriction in the capacity of a person to participate in or benefit from education”. Taking this broad definition, this study draws on the first longitudinal study of children in Ireland, Growing Up in Ireland, to generate a new estimate of SEN prevalence among Irish nine-year-old children. The analysis combines detailed information, collected from parents and teachers, encompassing diverse types of SEN, including physical disabilities, speech impairments, learning disabilities and emotional/behavioural difficulties. In doing so, the study establishes a SEN prevalence rate of 25 per cent among children in the mid-primary years, a rate very much in line with recent research in other European contexts. Additionally, the study details the diversity of data collected on children and young people with SEN and disabilities across agencies and government departments, the potential value of this data and directions for improved learner databases.

Background
There are few prospective studies of factors that mediate the association between exposure to adverse childhood experiences (ACEs) and obesity in adolescence. Our aim was to address this limitation.

Methods
We used prospective data from the Growing up in Ireland cohort study, with measurements at 9, 13, and 18 years old. The exposures were 14 adverse experiences before age 9. The main outcome was body mass index (BMI) at 18 years. Mediators were daily activity, diet quality, self-image and behavioural difficulties at 13 years.

Results
Among the 4561 adolescents in the final cohort, 77.2% experienced any adversity, 50.5% were female and 26.7% were overweight/obese at 18 years. BMI Z was higher at ages 9 (0.54 vs 0.43, p < 0.05, 95% CI of difference: −0.22, −0.01) and 13 years (0.50 vs 0.35, p < 0.05, 95% CI of difference: −0.25, −0.06), in those exposed to an ACE, compared to those unexposed. Structural equation models revealed that behavioural difficulties (β = 0.01; 95% CI: 0.007–0.018, p < 0.001) and self-concept (β = 0.0027; 95% CI: 0.0004–0.0050, p = 0.026) indirectly mediate the association between exposure to ACEs and BMI at 18 years.

Conclusions
The association between ACEs and BMI in adolescence is mediated by behavioural difficulties and self-concept.

Impact

• In a previous study, we found modest associations between exposure to a range of adverse childhood experiences and weight gain at 13 years of age.
• The strength of the association between adverse childhood experiences and weight gain was lower at 18 years of age compared to the association observed at 13 years and was no longer significant after controlling for confounding and including possible mediators.
• The association between adverse childhood experiences and BMI in adolescence is indirectly mediated by behavioural difficulties and self-concept.

Early nutrition plays a pivotal role in long-term health. The World Health Organization (WHO) recommends exclusive breastfeeding during the first six months of life, with the gradual introduction of solids after this period. However, studies in the Republic of Ireland (ROI) have shown poor compliance with guidelines. The ROI continues to have one of the lowest breastfeeding rates worldwide. Our objective was to analyse differences in breastfeeding and complimentary feeding behaviours between Irish and non-Irish mothers residing in the ROI, as well as the role of acculturation on these behaviours, using the national longitudinal study, Growing Up in Ireland (GUI). Mothers (n = 11,134) residing in the ROI were interviewed when their infants were nine months of age. The percentage of Irish mothers who initiated breastfeeding was 49.5%, as opposed to 88.1% among the non-Irish cohort (p < 0.001). Breastfeeding initiation reduced from 89.4% of non-Irish mothers who had arrived within the last year to five years ago to 67.5% for those who had arrived 11 to >20 years ago (p < 0.001). Our results indicate that cultural differences are an important factor in shaping patterns of infant feeding in the ROI. Reviewing existing support and education policies for parents is required to achieve the implementation of desirable infant feeding practices.

Keywords: infant feeding; breastfeeding; complimentary feeding; acculturation

Children’s parent and teacher relationships contribute to school adjustment and achievement, yet few studies have examined interactions between these relationships, particularly for father-child relationships. Using the Growing Up in Ireland birth cohort (N = 7,507 children, 50.3% male), we examined child-adult relationship quality – rated by mothers and fathers at age 3 and teachers at age 5 – as predictors of behavioural adjustment (rated by teachers) and academic achievement (formal reading assessments and self-reported academic self-concepts) at age 9. Controlling for prior levels of problem behaviours, verbal ability, and family SES, results indicated small and comparable independent effects of children’s parent and teacher relationships on school adjustment and achievement. For mothers and teachers, moderation analyses showed a cumulative risk pattern for conflictual relationships and a compensatory pattern for close relationships. Children are likely to benefit from improving closeness and reducing conflict in adult-child relationships and interventions that involve mothers, fathers, and teachers.

A recent rise in home computer ownership has seen a growing number of children using computers and accessing the internet from a younger age. This paper examines the link between children’s home computing and their academic performance in the areas of reading and mathematics. Data from the nine-year-old cohort of the Growing Up in Ireland survey is adopted for this study. The survey reveals that searching for information is a more popular activity among the children than communicating online. Through regression analysis we find that using a computer is positively and significantly associated with children’s reading and mathematics scores in standardised tests. This result holds after controlling for multiple determinants of school performance. In addition, we investigate the effects of using various applications on the computer. Surfing the internet for fun, doing projects for school and emailing are associated with higher reading and maths test scores, while those who are permitted to use the computer unsupervised tend to have higher maths test scores. Instant messaging and downloading music or watching movies are negatively associated with both reading and maths scores. The results indicate that some forms of early computer use have significant associations with academic performance among children in primary school, although we could not establish the direction of causation definitively.

Keywords: home computer use, elementary education, academic performance

Using data from Ireland’s national longitudinal study of children, this paper employs a capabilities approach to disability to understand how individual characteristics as well as home and school environmental factors at age 9 relate to academic and socio-emotional outcomes of students with special educational needs (SEN) at age 17. Results suggest that young people with SEN register both lower average scores and make less academic progress between the age of 9 and their national lower secondary examination, with the exception of young people with a physical SEN. Both home and school environmental factors at 9 years have long-term associations with the academic outcomes of young people with SEN, after controlling for individual characteristics and prior academic achievement. Home and school environmental factors had less consistent associations with the socio-emotional outcomes of young people with SEN. By using rigorous nationally representative longitudinal data, this paper offers a more holistic understanding of the development of young people with SEN. The paper also provides important evidence that a more inclusive approach for supporting students with additional needs, their parents, and their schools is needed.

Keywords
Special educational needs, capabilities approach, academic achievement, socio-emotional outcomes, Growing Up in Ireland

Parents’ and teachers’ beliefs and evaluations of young people are important. Using a feminist institutionalist perspective, and drawing on rich data from one in seven nine-year-old children in Ireland, this paper examines mothers’ (who make up the overwhelming majority of primary care-givers) and teachers’ perceptions of boys’ and girls’ mathematics performance. The evidence shows that girls’ mathematics performance is underestimated by both relative to boys’. Mother’s gender bias was evident among high performing children, at all levels of children’s academic self-concept, and among mothers with at least third level education. While the judgements reflect children’s actual performance and engagement, a notable gender gap remains. It is suggested that the results reflect gender stereotypes: overestimating boys’ and underestimating girls’ mathematics achievements. The article indicates the importance of the informal dimension of institutions and the part played by women in the effective devaluation of girls by endorsing gendered stereotypes. Women teachers are less likely to rate children highly in mathematics, taking account of performance: arguably reflecting their own lack of confidence in mathematics assessment. The findings raise concerns for girls’ futures since mathematics is seen as an indicator of intelligence. Given the move towards teacher-assessed grading during COVID-19, understanding, and challenging, gender-stereotyping is pressing.

Background
Unintended pregnancy is associated with increased risk for adverse neonatal and early childhood outcomes spanning an array of indicators, but it remains unclear whether these risks hold independent of other biological, social and environmental risk factors.

Methods
This study uses data from the first wave of the ‘Growing Up in Ireland Study’, a large nationally representative cohort study of more than 11 000 infants, to examine the risk factors associated with unintended pregnancy. Adopting a staged approach to the analysis, the study investigates whether pregnancy intention influences maternal health behaviours during pregnancy independent of background characteristics, and whether pregnancy intention carries any additional risk for adverse infant and maternal health outcomes when we adjust for background characteristics and prenatal behaviours.

Results
The study confirmed that sociodemographic factors are strongly associated with unintended pregnancy and that unintended pregnancy is associated with a range of health compromising behaviours that are known to be harmful to the developing fetus. While there was little evidence to suggest that pregnancy intention was associated with adverse neonatal outcomes or developmental delay independent of other covariates, there was strong evidence that intention status had a bearing on the mother’s psychosocial health. Unintended pregnancy was associated with increased risk of depression (risk ratio 1.36 [95% confidence interval 1.19, 1.54]), and higher parenting stress (risk ratio 1.27 [95% confidence interval 1.16, 1.38]).

Conclusions
Ascertaining the mother’s pregnancy intention during the first antenatal visit may represent a means for monitoring those at greatest risk for adverse mother and child outcomes.

Background
Social inequalities in the prevalence of childhood overweight and obesity are well-established, but less is known about when the social gradient first emerges and how it evolves across childhood and adolescence.

Objective
This study examines maternal education differentials in children’s body mass trajectories in infancy, childhood and adolescence using data from four contemporary European child cohorts.

Methods
Prospective data on children’s body mass index (BMI) were obtained from four cohort studies—Generation XXI (G21—Portugal), Growing Up in Ireland (GUI) infant and child cohorts, and the Millennium Cohort Study (MCS—UK)—involving a total sample of 41,399 children and 120,140 observations. Children’s BMI trajectories were modelled by maternal education level using mixed-effect models.

Results
Maternal educational inequalities in children’s BMI were evident as early as three years of age. Children from lower maternal educational backgrounds were characterised by accelerated BMI growth, and the extent of the disparity was such that boys from primary-educated backgrounds measured 0.42 kg/m2 (95% CI 0.24, 0.60) heavier at 7 years of age in G21, 0.90 kg/m2(95% CI 0.60, 1.19) heavier at 13 years of age in GUI and 0.75 kg/m2 (95% CI 0.52, 0.97) heavier in MCS at 14 years of age. The corresponding figures for girls were 0.71 kg/m2 (95% CI 0.50, 0.91), 1.31 kg/m2 (95% CI 1.00, 1.62) and 0.76 kg/m2 (95% CI 0.53, 1.00) in G21, GUI and MCS, respectively.

Conclusions
Maternal education is a strong predictor of BMI across European nations. Socio-economic differentials emerge early and widen across childhood, highlighting the need for early intervention.

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 AND OBJECTIVES
There is mixed evidence from correlational studies that breastfeeding impacts children’s development. Propensity score matching with large samples can be an effective tool to remove potential bias from observed confounders in correlational studies. The aim of this study was to investigate the impact of breastfeeding on children’s cognitive and noncognitive development at 3 and 5 years of age.

METHODS
Participants included ∼8000 families from the Growing Up in Ireland longitudinal infant cohort, who were identified from the Child Benefit Register and randomly selected to participate. Parent and teacher reports and standardized assessments were used to collect information on children’s problem behaviors, expressive vocabulary, and cognitive abilities at age 3 and 5 years. Breastfeeding information was collected via maternal report. Propensity score matching was used to compare the average treatment effects on those who were breastfed.

RESULTS
Before matching, breastfeeding was associated with better development on almost every outcome. After matching and adjustment for multiple testing, only 1 of the 13 outcomes remained statistically significant: children’s hyperactivity (difference score, –0.84; 95% confidence interval, –1.33 to –0.35) at age 3 years for children who were breastfed for at least 6 months. No statistically significant differences were observed postmatching on any outcome at age 5 years.

CONCLUSIONS
Although 1 positive benefit of breastfeeding was found by using propensity score matching, the effect size was modest in practical terms. No support was found for statistically significant gains at age 5 years, suggesting that the earlier observed benefit from breastfeeding may not be maintained once children enter school.

The Strengths and Difficulties Questionnaire (SDQ) is a brief 25-item instrument that has been widely employed in clinical and epidemiological studies to assess children’s psychological adjustment. Despite its widespread application in child and adolescent research, concerns have been expressed regarding the construct validity of the instrument and whether it might be tainted by a method factor that may undermine its utility as a diagnostic tool. We employed a confirmatory factor analytic approach to compare the goodness of fit of four competing models suggested by the extant literature for the parent-informant version of the questionnaire using data for 8514 nine-year-old children participating in the Growing Up in Ireland Study – a large population based cohort study in the Republic of Ireland. While analysis of the data provided support for the traditional five-factor conceptualisation of the instrument, a six-factor model which incorporated a method factor was found to fit the data marginally better. Nevertheless, we conclude that the existence of method effects does not present any great threat to the structural validity of the instrument taking account of patterns in the data and model parsimony.