Development of a predictive risk model for school readiness at age 3 years using the UK Millennium Cohort Study
Open Access
Abstract
Objectives The aim of this study is to develop a predictive risk model (PRM) for school readiness measured at age 3 years using perinatal and early infancy data. Design and participants This paper describes the development of a PRM. Predictors were identified from the UK Millennium Cohort Study wave 1 data, collected when participants were 9 months old. The outcome was school readiness at age 3 years, measured by the Bracken School Readiness Assessment. Stepwise selection and dominance analysis were used to specify two models. The models were compared by the area under the receiver operating characteristic curve (AUROC) and integrated discrimination improvement (IDI). Results Data were available for 9487 complete cases. At age 3, 11.7% (95% CI 11.0% to 12.3%) of children were not school ready. The variables identified were: parents’ Socio-Economic Classification, child’s ethnicity, maternal education, income band, sex, household number of children, mother’s age, low birth weight, mother’s mental health, infant developmental milestones, breastfeeding, parents’ employment, housing type. A parsimonious model included the first six listed variables (model 2). The AUROC for model 1 was 0.80 (95% CI 0.78 to 0.81) and 0.78 (95% CI 0.77 to 0.79) for model 2. Model 1 resulted in a small improvement in discrimination (IDI=1.3%, pConclusions Perinatal and infant risk factors predicted school readiness at age three with good discrimination. Social determinants were strong predictors of school readiness. This study demonstrates that school readiness can be predicted by six attributes collected around the time of birth.Keywords
Funding Information
- Medical Research Council
This publication has 33 references indexed in Scilit:
- Performance of logistic regression modeling: beyond the number of events per variable, the role of data structureJournal of Clinical Epidemiology, 2011
- Young Maternal Age and Poor Child Development: Predictive Validity From a Birth CohortPediatrics, 2011
- Multiple imputation using chained equations: Issues and guidance for practiceStatistics in Medicine, 2010
- Closing the gap in a generation: health equity through action on the social determinants of healthThe Lancet, 2008
- Infant developmental milestones and subsequent cognitive functionAnnals of Neurology, 2007
- Skill Formation and the Economics of Investing in Disadvantaged ChildrenScience, 2006
- The cost of dichotomising continuous variablesBMJ, 2006
- Internal validation of predictive modelsJournal of Clinical Epidemiology, 2001
- The assessment of parent-to-infant attachment: Development of a self-report questionnaire instrumentJournal of Reproductive and Infant Psychology, 1998
- The asymptotic distribution of the sum of weighted squared residuals in binary choice modelsStatistica Neerlandica, 1990