Background and rationale

This study will evaluate a programme that aims to embed social workers into schools (SWIS) so that they can work more effectively with education colleagues and with children and families. Education and Children’s Social Care (CSC) have an important inter-agency relationship, and both play a vital role in keeping children safe and promoting their wellbeing. Schools have long been among the major sources of referrals to CSC in the UK, contributing the second highest proportion (18%) of all referrals in 2018/19, behind the police (29%) [1]. Yet the two agencies have different roles and priorities, and significant cultural and organisational differences add to the complexity of working effectively together. SWIS was conceived before Covid-19, but the impact of the pandemic is also putting additional pressure on both schools and social workers [2, 3]. Policymakers have been increasingly interested in finding ways to improve how education and CSC work together to respond to safeguarding concerns and protect children, and in this context these efforts are likely to intensify.

SWIS may be a promising approach to doing this. The central idea is that having a social worker linked to and based within a secondary school can have a range of benefits. In particular, to improve the service delivered to children and families, enhance interagency working, reduce risks to children and lead to better outcomes. Building on a recent set of three pilot studies [4], this project is an evaluation of the next stage of development of SWIS: a scaling up trial involving 280 schools in 21 local authorities (LAs) across England, with half of these schools receiving a social worker. Aside from the SWIS pilots, there are several examples of social workers working in schools [5, 6]. However, the scale of the SWIS programme makes this the largest example of the approach.

The three pilot studies have been central to the development of the SWIS study. They took place in Southampton, Lambeth and Stockport in 2018–20. The pilot evaluation generated qualitative and limited quantitative evidence that the intervention was positively received by schools, social care, children and families [4]. A range of benefits were evidenced, including the opportunity for social workers to do more meaningful direct work with children, help a wider range of children and families than they otherwise would, and support schools to work through safeguarding issues. The evaluation also found indications of a reduction in Child in Need and Child Protection assessments carried out. However, the strength of the evidence was limited by the quasi-experimental design, small sample size, low incidence rates of key outcomes, and short follow-up. Taken together, the qualitative and quantitative findings from the pilots prompted the Department for Education to fund this larger trial of SWIS. Delivered through What Works for Children’s Social Care, this scale-up allows us to understand the potential effectiveness of SWIS more robustly.

Objectives

Study research questions (RQs) relate to (1) impact evaluation, (2) implementation and process evaluation, and (3) economic evaluation.

Trial design

The study is a pragmatic cluster randomised controlled trial with two arms (one intervention and one control group) and mainstream secondary schools as the unit of randomisation.

Study setting

The study will be conducted in mainstream secondary schools in England, across 21 LA areas.

Eligibility criteria

Mainstream secondary schools are places of education for children aged between 11 and 18 (Years 7 to 13). They are funded by the Government and provide a free education for children, though a number of models exist, such as academies, free schools, and faith schools. All children attending the schools are eligible for the trial. Schools may opt-out of participation in the Implementation and Process Evaluation (IPE) while remaining in the trial.

Who will take informed consent?

Informed consent will be obtained from all individuals who take part in aspects of the IPE (e.g. surveys and interviews). Researchers will take consent, and receive signed consent forms and/ or audio recorded verbal consent from participants.

Intervention and control conditions

The intervention physically locates social workers within schools with the aim to build better working relationships with school staff, students and families. Rather than working with students and families from a LA office base, and liaising with and providing advice to education professionals remotely, the social worker is embedded in the school. S1 Appendix shows a logic model that describes three hypothesized mechanisms of impact:

1. Enhanced school response to safeguarding issues
2. Increased collaboration between social worker and school staff, and parents
3. Improved relationships between social worker and young people

Pathway A is based around regular, good quality, communication between the social worker and school staff. The advice and support given to school staff is hypothesized to increase staff confidence in safeguarding issues and improve the quality of school referrals. Pathway B is about working directly with families and improving relationships between social workers and parents. Pathway C is about working with children and young people directly. Frequent interactions with the social worker enable the young person to trust the social worker and to feel understood and supported. This is theorised to lead to improved school attendance and participation, better management of a young person’s risks and improved outcomes. In all three pathways, improved child and family outcomes are theorized to lead to a reduction of the number of children in care.

Following the pilot evaluations, an intervention manual has been developed (WWCSC, 2020 –available on request from programmes@whatworks-csc.org.uk). This distils the key messages from the pilots into a practical format designed to assist the 21 LAs in delivering SWIS. Its stated aim is to provide “a framework to refine the SWIS Scale-up programme and encourage more consistent and effective integration of social workers into schools”, although the need for flexibility is acknowledged. The manual offers a number of recommendations for implementers. Among them, the following relate to programme delivery:

• Social workers should be embedded within secondary schools, but can work with feeder primaries.
• Social workers linked to schools should be experienced (being in practice for at least 2 years).
• The focus should be on statutory social work with additional opportunities for “preventative” aspects, which could involve “advising staff, families and young people” and working with children who are not at the threshold for formal involvement.
• Caseloads should be managed by the SWIS team manager and be in line with LA averages, and where possible the carry-over of existing caseloads prior to the launch should be minimal. To avoid disrupting existing relationships SWIS workers are expected to take on new cases.
• Social workers should be embedded in schools as far as possible, with their own office space in the school and opportunities to integrate.
• Face to face contact should be the basis for the intervention, in order to build strong relationships with school staff, children and families.

The control group will receive service as usual. Children who are deemed by school staff to require the involvement of CSC will be referred to the LA, usually via telephone call or email to a Multi-Agency Safeguarding Hub arrangement or a referral and assessment team. Children judged by CSC to meet the threshold for involvement will be allocated a social worker as usual, but social workers will not be based in the school. This was chosen as the comparator because it is the status quo that the intervention can be distinguished from, and the system that SWIS needs to perform better than if it is to be considered worthwhile. Children who are deemed to require the involvement of CSC will continue to be eligible for this across all schools, including the control group. LAs will make case by case judgements about the best interests of children who were receiving a service prior to the trial. For example, a child on an existing child protection plan may retain their previously allocated social worker to avoid disruption, rather than transferring to the SWIS worker.

Strategies to improve adherence to interventions

We anticipate good adherence to the intervention, as the funding for the trial intervention (i.e. the social worker) is provided to participating LAs and schools on the basis of compliance with the intervention guide supplied by the funder, and LAs have committed to delivering SWIS as described. All schools also confirmed their commitment at the outset. However, the adverse circumstances of Covid-19 may mean that some sites need to adapt SWIS, for example to accommodate social distancing or increased remote working. The IPE is designed to capture such adaptation and secondary analyses will explore the impact of varied levels of adherence.

Outcomes

Sample size

At the study design stage, the funder advised that a minimum of 280 mainstream schools would be available to be randomised. Assuming an average of 925 students per school, an average base rate of 12.6 s.47 enquiries per 1000 students per school year under usual practice conditions, and a between-school coefficient of variation of 0.45 within LAs (these estimates are all based on comparator school data from the three pilot studies in Lambeth, Stockport and Southampton) [4], randomising 140 mainstream schools to each group provides 90% power to detect a decrease in rates from 12.6 to 10.48 per 1000 pupils per school year (i.e. a rate ratio of 83.2%). This is using a two-sided 5% type I error level when using a Poisson regression model accounting for cluster randomisation [8].

Recruitment

Participating LAs were chosen through a competitive tender process managed by the funder. Each chosen LA invited schools and gained the agreement from up to 16 schools to be put forward for randomisation. Recruitment of schools was completed for each LA before the list of schools was passed on to the trial statistician for randomisation. The statistician was not directly involved in the recruitment of schools. Schools were considered recruited once the LA confirmed that they have agreed to take part.

Allocation to intervention or control groups

At the beginning of the study the research team received a list of up to 16 schools which had agreed to take part from each LA. The trial statistician performed the randomisation as described below, and the research team communicated the allocations to LAs via the funder.

Sequence generation

Mainstream schools were allocated to SWIS or usual practice in a 1:1 ratio whilst minimising covariate imbalance within and across blocks using a balancing algorithm for cluster randomised trials with multiple blocks [9], with LAs acting as blocks. This was implemented in R version 3.6.0 [10] using code provided as supplementary material to [9]. For the first block, the standard imbalance metric (Equation 1 in [9]) was used. The allocation of subsequent blocks was conditional on blocks already allocated, using a modified imbalance metric (Equation 2 in [9]). Balancing variables were school size (total number of students enrolled in year 7 and upward) and proportion of students eligible for free school meals (FSM). Both balancing variables were weighted equally, and will be adjusted for in the final statistical analysis. The rationale for selecting these variables is as follows:

• School size (total number of students enrolled). The size of the school and number of students is likely to have an effect on how the social worker engages with and works within the school. In a larger school their time and resources may be spread across more students, and there may be more professionals to work with. It is reasonable to expect that this is a factor that will shape the implementation of SWIS and therefore balancing school size between groups is sensible.
• Proportion of students eligible for free school meals. FSM are provided for children resident in lower income households. They are designed to ensure that all children have access to adequate nutrition when they are at school. Eligibility for FSM is a reliable indicator that a child is from a low income household [11]. This is important because the work of CSC tends to be focussed on children from lower income households, and it is these children who are more likely to require a service from CSC [12].

Data collection

Data management

The following management plans are in place for each type of data.

Confidentiality and data security

Personal data will be transferred to Cardiff University for the IPE. Personal data (name and contact details) of social workers, school staff and children will be transferred to Cardiff University by LAs for those who have agreed for this information to be sent. Study information will be made available to IPE participants prior to participation. A risk assessment has been signed off to mitigate any potential risks regarding confidentiality and identification while processing these data, and personal data will be stored on secure university network servers, with access limited to members of the research team. No personal data will be processed by Cardiff University for the impact evaluation or economic workstream. All data will be stored in a secure manner and processed in accordance with data protection legislation (in accordance to GDPR and UK DPA 18) and Good Clinical Practice (GCP).

The elements and timing of data collection are shown in Fig 1 below.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. SPIRIT checklist.

https://doi.org/10.1371/journal.pone.0265354.g001

Statistical methods

Ethical considerations

The study has ethical approval from School of Social Sciences Research Ethics Committee of Cardiff University (Ref: SREC/3865, approval date: 26.08.2020). Written, informed consent to participate will be obtained from all participants. Due to the use of remote methods, and in case consent forms are not returned, verbal consent will be audio recorded in interviews with professionals in addition to written consent.

If interviewees say anything that makes the researcher concerned about harm to the participant or another person, then they have a duty to take appropriate action. In the first instance, usually this would involve discussing the concern with the Chief Investigator or a co-investigator. Depending on the nature of the harm, referrals to agencies may be appropriate, for example a referral to the LA CSC may be deemed necessary if a child was thought to be at risk.

Trial status

LAs are classed as ‘recruited’ once they have signed the data sharing agreement. Recruitment began (data sharing agreements sent for signature) on 22 October 2020. Recruitment was completed in February 2021. The trial was extended twice by the funder, first in August 2021 and then in January 2022 (see below) Amended data sharing agreements to cover the extended period were circulated to LAs for signature n September 2021 and March 2022.

Amendments to the study

The study was extended twice by the funder, first in August 2021, and then in January 2022.This protocol was amended to accommodate the new timeline. This document reflects the amended study design. Both the original protocol (version 1) and the revised protocol (versions 2 and 3) can be found on the OSF. Key amendments made in the updated protocols are:

• Timing of analysis and reporting updated to reflect the extended intervention period (e.g. meaning that the main impact analysis will now be published in January 2023 rather than 2021, and the follow-up analysis and final report will now be published in January 2024 instead of June 2022).
• Changes to the data collection activities in the IPE, to reflect the fact that all case studies were forced online by the pandemic and conducted remotely. This meant that as of September 2021 no observations or child interviews have taken place, and these are now scheduled for terms 4 and 5.
• Additional data collection activities in the three extra terms, e.g. interviews with key decision makers in LA ‘front door’ teams.
• The use of the 2018/19 school year as a baseline, rather than the 2019/20 school year as originally planned. This is because the 2018/19 year is the most recent year not affected by the Covid-19 pandemic and therefore a more realistic baseline to use.
• Changes to the economic evaluation to reflect discounting to costs beyond 12 months from the start of the intervention, to present values using nationally recommended discount rates.

Dissemination plans

We are contracted to report the findings in reports that will be published open access by the funder at https://whatworks-csc.org.uk/. We will also disseminate the study widely by other means, using in-person methods (e.g. conference presentations and invited talks), electronic methods (e.g. taking part in a podcast and online videos produced by the funder, publishing blogs on the university webpages) and by publishing in academic journals.