Using Childhood Activity and Sleep to Predict Emotional and Behavioural Difficulties at Age 11

Abstract

This study utilised data from the Auckland Birthweight Collaborative (ABC) longitudinal study. The ABC study was established to compare the developmental trajectories of full-term children born small for their gestational age (SGA), with their average birthweight peers (appropriate for gestational age; AGA). Data has been collected at birth, and at the ages of 1, 3.5, 7, and 11. At each phase extensive amounts of physical, psychological, developmental, and familial data have been collected from the children and their families. Central to the current study, at the ages of 7 and 11 the children wore an Actigraph monitor, providing an objective measure of their daytime activity and sleep duration. At each age, two childhood behavioural questionnaires were also administered, one used commonly to screen for Attention Deficit Hyperactivity Disorder (ADHD), and the other for more general psychological difficulties; the Conners’ Rating Scale (CRS) and the Strengths and Difficulties Questionnaire (SDQ) respectively. Data was collected successfully for 540 children at both ages across these measures. Changes in activity and sleep are acknowledged to be indicators of over 30 psychological disorders. This study sought to examine whether objective measures of activity and sleep duration can be used to predict emotional and behavioural difficulties in 11 year old children, as measured by the CRS and the SDQ. This was with the overall aim of exploring whether activity and sleep duration measures may have diagnostic utility beyond more established risk factors. Given that accelerometers in smart phones now offer the opportunity to measure these variables in real world settings, such findings would suggest that these technologies could be used to inform clinical psychological assessment. It was also hoped to examine how accurately symptoms of motoric hyperactivity are identified by subjective behavioural measures. The literature review explores the role of the two questionnaires in a New Zealand context, and the technical feasibility of using activity and sleep measures diagnostically. It first identifies the central and stable contribution of activity to a child’s temperament, noting that positive associations with activity reverse as children get older. While concern is often focussed on the excessive activity associated with externalising disorders, it is found that activity is also a valuable predictor of internalising disorders. The evolving nature of the ADHD diagnosis is explored, noting the reduced emphasis placed on hyperactivity in recent years. Broader controversies around the diagnoses are then introduced, highlighting an arguably premature move away from the consideration of psychosocial aetiological factors. An overview of the research techniques used to measure activity and sleep is then presented, including how well these measures can distinguish subtypes of ADHD, conduct disorder, and internalising disorders. A more detailed consideration is given to why the role of genetics in ADHD may have been overestimated at the expense of environmental factors, before the environmental and psychosocial covariates utilised in this study are introduced. Data screening highlighted that compared to the original ABC sample at birth, the children who provided data at age 11 were likely to have: had a higher birthweight; been born to older parents with a higher family income and who were married; and to mothers who reported less stress, and were non-smokers. In addition, due to participant attrition, the sample was restricted at an earlier stage of the study to European mothers. As such, some caution must be employed in generalising the results. Unsurprisingly, given that clinical screening questionnaires were being used on a general population, the outcome measures were also heavily skewed. It had been hoped to focus the study on predictors of ADHD, examining the predictive utility of the subscales of individual outcome measures, and contrasting Teacher and Parental ratings of hyperactivity. Due to normality restrictions, however, the broader SDQ Total Difficulties score (continuous and categorical), and the CRS ADHD Index score (categorical), were selected as outcome measures. This thesis is divided into five main stages of analysis. Firstly, in the absence of age specific norms, the appraisal of developmentally inappropriate activity and sleep is difficult. This research therefore aimed to provide such normative data in a New Zealand context, providing summaries of activity measures and sleep duration, as well as SDQ and CRS scores, at the age of 11. The data was also examined for main effects of gender and birthweight using 2x2 MANOVAs. Secondly, the years between the ages of 7 and 11 represent a key developmental period for children, during which behavioural expectations increase greatly; raising the likelihood of developmental difficulties being exposed. As a central objective of the ABC study, all measures were examined between the ages of 7 and 11 using repeated-measures MANOVAs, to see if there had been differential changes across time between AGA and SGA children. Thirdly, univariate correlational analyses were conducted to explore the relationship between activity and sleep duration, and the outcome measures. A variety of activity measures were employed to identify which has the most utility as a predictor. Analyses were also conducted with 48 more established risk predictors. Predictors showing significant correlations (p<.01; p<.05 for activity and sleep) were then examined for multicolinearity, before being taken forward for further analysis. Fourthly, predictors that showed significant independent relationships as predictors were modelled using hierarchical regression analyses. Finally, to explore the central hypotheses of this thesis, exploratory univariate analyse, using linear and logistical regression, were conducted on the Internalising and Externalising subscales of the SDQ and the CRS ADHD subscales. The study therefore offers a valuable normative sample of 11 year old New Zealand European children’s results on the CRS and SDQ, as well as their activity levels and sleep duration. For the first time in the ABC study’s history the current research showed SGA children to perform more poorly on certain psychological measures. Examination of these measures between the ages of 7 and 11 indicated the possibility that previously latent difficulties are being uncovered as SGA children grow older. While the lack of normality of the subscale scores prevented definitive conclusions, internalising emotional difficulties were more evident among SGA females, but SGA males exhibited broader issues across both the internalising and externalising domains. An examination is then presented of the relationship between Parental CRS and SDQ scores, activity and sleep duration, and established risk factors. These risk factors include: low-birthweight; prenatal alcohol, nicotine, and cannabis exposure; psychosocial adversity; maternal social support and perceived stress; child intellectual function; and child electronic media exposure. These analyses found theoretically consistent relationships between activity and the outcome variables, but activity measures were outperformed by the covariates as predictors. The most notable of these predictors were maternal stress, maternal cannabis use, and child IQ. Sleep duration was also a very poor predictor and neither variable appeared to have diagnostic utility based on the measures employed in this study. Environmental and psychosocial variables accounted for significant percentages of variance of the outcome measures, reemphasising the importance of these oft neglected factors. While normality restrictions again limited the conclusions that could be drawn, the strength of associations between activity and Parental CRS ADHD subscale scores suggest these scales capture motoric hyperactivity, but that teacher scales may capture the relationship between low activity, sleep duration, and inattention.