What factors are associated with children being taken into care by the state after initial contact with services? A survival analysis of Children’s Social Care data in Liverpool

Abstract

Background Increasing numbers of children in the UK are being taken into care, with adverse consequences for children and unsustainable costs for local government. It is crucial that local authorities better understand which children are most at risk to target preventative interventions.Objectives To identify predictors of children becoming ‘looked after’ (taken into care by the state) among children known to a local authority.Methods Secondary analysis of routinely collected Children’s Social Care data including all children who first became known to social care in Liverpool between April 2019 and March 2022, (excluding unaccompanied asylum seeker children). Outcome was time (in months) between first contact with social care and either becoming looked after or reaching the censoring date (March 2022). Survival analysis was undertaken using a discrete time hazard model.Results 5808 children under 19 became known to Children’s Social Care during the study, 377 of whom subsequently became looked after. Black and Asian children were more likely to become looked after, as were those known to social care services pre-birth or in the first year. Key risk factors that increased hazard of becoming looked after were neglect, sexual abuse, emotional abuse, drug and/or alcohol use in the household or the child and mental ill health in the household. Children who had a Child in Need intervention were less likely to become looked after. Children who had a Child Protection Plan were more likely to become looked after.Conclusion In our study of routinely collected Children’s Social Care data, we have identified the key factors that increased the hazard of a child becoming looked after. These findings highlight potential areas for service change and can be used to inform risk prediction and preventative action, however, the local context may influence the generalisability of our findings to other settings.