OP14 How well can early child development be predicted in children in the UK? findings from the millennium cohort study. (September 2018)

• Record Type: Journal Article
• Authors: Camacho, CL; Straatmann, VS; Day, JC; Taylor-Robinson, DC
• Is Part Of: Journal of epidemiology and community health. Volume 72(2018)Supplement 1
• Page Start: A7
• Page End: A7
• Publication Date: 2018-09
• Subjects: school readiness -- predictive risk model -- social determinants; Public health -- Periodicals; Epidemiology -- Periodicals; 614.4
• DOI: 10.1136/jech-2018-SSMabstracts.14
• Languages: English
• ISSNs: 0143-005X

Abstract

Abstract : Background: Early cognitive development (ECD) is an important determinant of educational attainment and health outcomes. ECD is socially patterned and influenced by biological, environmental and socioeconomic factors. This study aimed to identify how well socio-economic and demographic factors from the first year of life predicted ECD at age 3 in a contemporary UK cohort. Methods: We used data on 9487 complete cases children from the UK Millennium Cohort Study collected at 9 months and 3 years old. The outcome was school readiness using the Bracken School Readiness Assessment (BSRA) assessed at age 3 years. Predictive risk modelling (PRM) was carried out and the discriminatory capacity of two models was compared by the area under the receiver operating characteristic curve (AUROC; 95% Confidence Intervals). Stepwise statistical selection specified a model with 13 perinatal and sociodemographic predictors collected at age 9 months. This was compared with a parsimonious model comprising the top 6 predictors, identified by dominance analyses which ranks the importance of each variable in the model. Integrated discrimination improvement (IDI) was also calculated, comparing the two models. All analyses were conducted in Stata SE version 14.2. Results: At age 3, 11.7% (11.0%–12.3%) of children were not school ready. The 13 variables statistically selected were: parents' National Statistics Socio-Economic Classification, child's ethnic group, maternal education, income band, sex, number of children in the household, mother's age, low birth weight, mother's mental health, infant developmental milestones, breastfeeding, parents' employment, housing type. The parsimonious model included the first six listed variables, based on dominance analyses. The AUROC for the full model (13 predictors) was 0.80 (0.78–0.81) and 0.78 (0.77–0.79) for the reduced model (6 predictors). IDI showed there was a small but significant difference in performance, with the full model resulting in a 1.3% (p≤0.001) improvement in discrimination. Conclusion: We identified a set of predictive risk factors from the perinatal period and early infancy that predicted school readiness at age 3 with good discrimination. Social factors were the strongest predictors of school readiness. This study demonstrates the feasibility of predicting school readiness using just six attributes collected around the time of birth. PRMs could be used to identify children who would potentially benefit most from early interventions. Further research to assess the impact of this PRM would be required before it could be used in practice. A strength is the nationally representative sample; a limitation is the lack of external validation.