This project seeks to evaluate the roll-out of the Family Safeguarding model in five local authorities in England, as part of the Department for Education (DfE) funded Strengthening Families, Protecting Children (SFPC) programme. The local authorities analysed are Walsall, Lancashire, Telford & Wrekin, Wandsworth and Swindon.
The evaluation of the roll-out includes an impact evaluation of the model, with a stepped wedge cluster Randomised Controlled Trial (RCT) design. This is accompanied by an Implementation and Process Evaluation (IPE) across the same five local authorities, to understand the delivery during the rollout of the model.
The Family Safeguarding Model
Developed in Hertfordshire, Family Safeguarding is a model of system change where specialist adult practitioners are embedded within Family Safeguarding teams of social workers. The intervention supports a whole-system change to a local authority’s child protection approach for children up to adolescence, focusing on supporting the needs of children and adults in order that children can safely remain within their families. This involves:
- Establishing multi-disciplinary teams where specialist adult practitioners in domestic abuse, mental health and substance misuse are co-located with social workers under a unified management structure. This enables a multi-disciplinary whole family response through direct assessment and support from specialist adult practitioners as well as multi-professional group case discussions and sharing of knowledge and skills across disciplines.
- Use of Motivational Interviewing (MI) as a framework for practice for all staff. Staff undergo training and ongoing skills development workshops and follow a structured solution-focussed intervention programme with families which aims to work collaboratively with families and increase engagement.
- Using an electronic assessment workbook which provides a single data tool for all professionals and links to the work programme. This increases ease of information sharing between professionals and reduces social worker time spent recording and sharing information.
The study design of the impact evaluation is a cross-sectional stepped-wedge cluster randomised controlled trial, where the timing of implementation is staggered across local authorities. The point at which local authorities begin implementing the intervention is selected at random, constrained by their level of readiness to implement the model. In this way, all the local authorities in the sample will eventually implement the Family Safeguarding model, but randomising the start date of the implementation of Family Safeguarding will allow service users in the local authorities that have not yet implemented the programme to act as a control group against service in local authorities where Family Safeguarding has already been implemented.
The primary evaluation question of the stepped wedge RCT relates to care status: What is the impact of Family Safeguarding on the likelihood of children becoming looked after?
There are also five secondary questions, relating to changes in: i) the likelihood of returning to statutory services after a child protection plan (CPP), ii) the time spent on CPPs, iii) the likelihood of a case progressing from pre-proceedings (under the PLO) to care proceedings, iv) the likelihood of children being re-referred for parental substance misuse, domestic violence or parental mental health issues, and v) unauthorised school absence rates of children referred to children’s social care.
Given the challenges the COVID-19 pandemic poses to evaluating a stepped-wedge RCT, a difference-in-differences analysis will be conducted in addition to provide an additional approach to analysing the programmes’ effects and to increase the robustness of the impact evaluation estimates. The analytical approach for the difference-in-differences analysis is set out in a separate protocol at the end of the page.
Implementation and Process Evaluation
The purpose of this implementation and process evaluation is to assess delivery during the rollout of the Family Safeguarding Model across five local authorities. The aim of this is to help understand and explain any identified intervention effects (or lack thereof) in the concurrent stepped-wedge cluster randomised controlled trial, to identify elements of successful delivery, and to improve the understanding of the model.
The implementation and process evaluation seeks to answer the following research questions:
- Fidelity and adaptation
- To what extent does delivery in participating authorities adhere to the model?
- Are the key assumptions and facilitating factors in place?
- Programme differentiation
- What does the existing service structure and practice look like in participating authorities prior to the introduction of the model?
- Reach and acceptability
- What is the number and characteristics of families reached by the intervention?
- What is the experience of staff and families who have been involved with the intervention?
- Does implementing the model lead to perceived changes in the interim and ultimate outcomes identified in the logic model?
- Is the level of effectiveness of the model perceived to differ for different groups?
- Are there any perceived unintended or negative consequences as a result of introducing the intervention?
The first local authority is expected to begin implementing Family Safeguarding in the Spring of 2020, with the other local authorities following in 6-month intervals. Final outcome data is planned to be collected in October 2025, with final reporting expected in early 2026.
We anticipate that due to the COVID-19 pandemic that it is likely there will be delays to these timings.
The COVID-19 pandemic, along with associated changes to programme delivery, are likely to undermine our original stepped wedge design. As such, we are adding a supplementary matched difference-in-differences (DiD) analysis to the current research design. The changes to programme delivery have also delayed some of the data collection for our implementation and process evaluation, but this eventuality was already accounted for in our original trial protocols so no significant changes will be required.