What could be done differently? The evidence above indicates an opportunity to better connect families to specialist local authority support, when they need it. It also suggests that a potential solution could be to make better use of existing, wider system interactions with families to identify the need for this support. This is not aimed at ensuring that all at-risk children avoid coming into care – for many, coming in to care will remain the right outcome. Instead, the aim is that no family in need should reach crisis point without ever having been connected to available support. Practitioners already exercise considerable skill and judgment in making decisions in time-, resource- and information-constrained environments. The goal of better data infrastructure is to give them a more complete picture to work with, not to replace the relational practice that sits at the heart of good work with families. There are existing systems in place which should mean that practitioners have a single view of intelligence about a child or family, that draws on multiple sources such as health, police, schools or local authorities. However, the existing systems can be labour intensive, time consuming and sometimes unreliable. Furthermore, they tend to drive reactive behaviour, as practitioners see families already flagged as being of concern, and will make decisions about what support they might need. In contrast, in an environment in which practitioners have access to wider data, individual families that could benefit from targeted support are identified before needs escalate, and practitioners can operate more proactively. I always had to re‑share really triggering information with people who should have been working together and already aware of my situation.” Care experienced young person It took a placement review for them to realise I hadn’t been in education — it should never have taken that long.” Care experienced young person
This could then work alongside the changes described in shifts one to three, whereby this data led intelligence is combined with ‘on the ground’ understanding of wider practitioners and professionals in the system (in adults services, schools or as Family Help Lead Practitioners) to successfully engage the families that most need support earlier.
If services could make appropriate use of a comprehensive connected dataset, across the whole local partnership, their ability to identify families requiring support could be significantly enhanced, and in some instances, an earlier, targeted decision to support a child or family might be made as a consequence.
Case study: One council has been able to analytically identify approximately half of the children and young people who would otherwise enter care without a prior Early Help/Child In Need/Child Protection plan, and are now in the process of determining how to best support those families. This was done by using pre-existing data sharing agreements in the partnership, that permit data sharing between partners in the interests of safeguarding children (commonplace across many local authorities). Resident-level data sets across children’s social care, education, housing, revenue and benefits and adult social care were combined to create single family records across multiple services. A machine learning model was then run across three years of data to determine the risk events that correlate most strongly with the outcome of a child entering care on this pathway, essentially to predict the likelihood of a child entering care. Developing the model took approximately three months after the initial data collection exercise across partners. In that three-month period, approximately half of the children that did enter care without prior EH/CIN/CP plan were highlighted by the model.
The risk events identified based on this authority’s data were: rolling 6 month contact frequency, education related flags (such as truancy or persistent exclusion), housing instability. This approach identified a list of families where those risk events had occurred but had not yet received support from the local authority’s children’s social care department, and who were therefore high priority for support. The council is now turning its attention to running the model continuously so that the families who could benefit from early support are identified in real time, as well as designing how to operationally organise partners around these families so that they get the right support. It is important to note that tools of this kind are designed to support and not replace professional judgment. Analytical identification of families at potential risk should always be followed by skilled, relational practice from experienced practitioners who can assess the full picture of a family's circumstances. Algorithmic flags are a prompt for human attention, not a basis for intervention in their own right. The ethical and practice governance frameworks around how such tools are used, and how families are approached as a result, require careful design.
These findings suggest that for some children, earlier identification and connection to support may have been possible. Earlier identification does not automatically translate into different outcomes – entry to care often follows sustained work and complex risk and is frequently the right decision for a child’s safety. The opportunity here is to ensure families are connected to the right support when need is present, not to suggest that earlier contact would definitively have prevented care in all or most cases. Summary: This evidence suggests that there appears to be potential to provide targeted support to 74% of children in care at an earlier stage, rather than them first receiving support at Child Protection Plan stage or coming in to care without immediate prior support through an Early Help or Child in Need plan. It also indicates that the information needed to identify families at risk does often exist across partner organisations, but could be joined up more often to enable prompt earlier support from local authority safeguarding specialists.
48
49
Made with FlippingBook - Online catalogs