Case Study: Understanding Missing From Care Episodes - BCFT and LIDA

Aims and Objectives

Bradford Children and Families Trust approached HDRC Bradford and LIDA (Leeds Institute of Data Analytics) to help analyse the data associated with Looked After Children (LAC). Looked after children are among the most vulnerable young people in our communities, and when they go missing, the risks escalate sharply. Yet despite the scale of the issue, services still lack clear, data driven insight into who is most at risk, when that risk is greatest, and what factors drive repeated episodes.

HDRC Bradford linked BCFT to LIDA and its Data Science Development Programme (DSDP), a programme dedicated to training data scientists using real-world data and partners.

BCFT already shares routine data with the Connected Bradford programme, so HDRC Bradford ensured the data scientists had access to the platform and accurate up-to-date data, as well as introducing the research team to BCFT’s decision-makers and analysts. Working with Lewis Shaw (LIDA), Prof. Toby Davies (University of Leeds), Dr Megan Wood (Bradford Institute of Health Research) and Prof. Daniel Birks (University of Leeds), we aimed to create the clearest and most robust picture to date of missing from care episodes in Bradford. The project received support from Rob Shore, HDRC Data Manager for his expertise in routine data linkage and context and HDRC Data Scientist Dr Yanhua Xu to mentor the young researchers in data science methodologies and approaches.  
 
Our objectives were to:

• Understand patterns and concentration of missing episodes across all looked after children.
• Identify specific risk factors - individual, environmental and educational - that influence whether a child goes missing.
• Analyse the timing of episodes to pinpoint critical risk windows, especially around placement changes and returns from missing incidents.
• Equip practitioners with insights that can shape prevention, early intervention and safer care planning.

The Work 

Using the Connected Bradford Research Database, we brought together:

• Data on 2,900 looked after children (2018–2025).
• Over 5,800 missing episodes linked to 5,500 placements, each mapped to the placement active at the time.
• Education records (absence, exclusion, SEN / SEND, EHCPs) from the Department for Education to explore wider vulnerabilities.

This integrated dataset enabled us to examine missing episodes not just as incidents, but as part of the broader life circumstances of each young person. The collaborative research team met weekly and worked iteratively the develop the approach with BCFT’s data analysts and practitioners throughout, to aid real-world understanding and context of the data. BCFT was keen to understand risk, not just counts. As a result, the data scientists used survival analysis. This is the same powerful statistical methods used in public health to understand when adverse events (like hospital readmissions) are most likely to occur.

The team at the University of Leeds modelled daily missing-episode risk using a Cox proportional hazard approach to assess how factors like age, placement type, prior protection plans and school engagement shape the timing and likelihood of first and repeat missing episodes. Missing episodes are highly concentrated in a small group of children, with risk driven more by placement environment than individual traits, amplified for older teens and strongly indicated by persistent school absence. The research team established that the first 20 days of a placement are a high-risk window, with risk quadrupling immediately after a missing episode and then shifting over time.

Report: Understanding Missing from Care Episodes

Impact

The Bradford HDRC Data Science team and LIDA researchers have been able to turn years of untapped routine data on children missing from care data into real insight. Previously, data was mostly used for KPI reporting, meaning services knew who went missing most, but not why or when risk intensified. This work changed that. The project bridged the gap between academic and local government and pinpointed the small group of children at highest risk, identified the most critical periods, and showed that environment, not individual traits, drives risk most.

BCFT is keen to use these findings to shape strategic planning, commissioning and shared understanding. The findings will drive stronger early intervention in the first weeks of a placement, better support after a missing child returns, and smarter placement matching and commissioning.
This analysis could also allow services to prioritise the small group of children at highest risk and give all agencies a shared evidence base for more consistent safeguarding. Overall, the work creates a clearer, more coordinated approach to preventing missing episodes.

What next?

BCFT plans to use these findings, supported by linked local authority data on placement history, safeguarding incidents and risk assessments, to inform targeted interventions, improve safeguarding practice and develop proactive approaches for children most vulnerable to going missing. The work also provides a model that HDRCs can replicate using their own linked datasets and aligns with national policy priorities around evidence based early intervention, integrated services and ethical, transparent use of predictive analytics in children’s social care.