Introduction

Family group conferences (FGCs) were first introduced in the 1980s to reduce the numbers of Māori children in state care in Aotearoa New Zealand (Ban 2005). Since then, they have spread across much of the world and are still employed predominantly to prevent children coming into care or to return them home from care (Dijkstra et al. 2016; McGinn et al. 2020; Nurmatov et al. 2020). They are widely used in the UK with 79 percent of local authorities (LAs) reporting that they offer them (Wood et al. 2024). Their use is likely to increase, given the Children’s Wellbeing and Schools Bill mandates family group decision-making for children in England when a court order is being seriously considered to bring a child into state care. Wood et al.'s (2024) survey found the use of FGCs in the UK has expanded beyond keeping children out of care. Most LAs offer early help FGCs, while many also offer them to support children who are in state care, for example, to (re) build relationships with birth families and prepare for leaving care. Some LAs also use them in adult services (Tew et al. 2025). There is little evidence on how the experiences of families having FGCs, and their outcomes over time, compare across different stages of child welfare involvement. This study aims to help fill that gap.

Background

The FGC model and the UK child welfare context

The idea of an FGC is to move decision-making about a child (or children) in need or at risk from professionals to the wider family. In many countries with mature child protection systems, professionals dominate the processes in place for keeping children safe (Ney, Stoltz, and Maloney 2013). The FGC model, as quality assured in the UK by the Family Rights Group (2026), involves a coordinator, who is a family support professional, but independent of the social work team with case responsibility, meeting with primary carers to start the process of convening a meeting of wider family members and any key family friends, as identified by the primary carers. This network will then be invited to attend a family meeting/conference to discuss the best way to look after the child who is in some kind of need, including how any or all family members can offer support. If there is a case-holding social worker, they will usually be present for the initial discussion, but then they, along with the coordinator, will leave the family alone for ‘private family time’. If there have been child protection concerns, the family-led plan is then presented to the local statutory social services who will decide if it is viable. There are other variants of family group decision making, although some (e.g. Merkel-Holguin 2004) argue the FGC is the most participative model. One variation of FGCs is a shuttle model, where there is particularly difficult conflict between parents or a history of domestic abuse. This allows for two parallel discussions facilitated by the same coordinator.

The UK child welfare system is described, in international comparisons, as having a child protection orientation, rather than a family service orientation (Gilbert, Parton, and Skivenes 2011). The UK’s four nations have devolved responsibility for social services, but their child welfare systems are substantially very similar. The service terminology used in each nation is slightly different, but there is a common division according to the stage of concern about a child–early help/intervention which is preventative; statutory support, beginning with voluntary ‘child in need’ (or equivalent) services; a formal child protection plan where a child is thought to be at risk of significant harm; then care proceedings if a child needs to be legally placed in the care of the state, which usually means them being removed from their birth family.

Realist evaluation

The implementation context of any social intervention can fundamentally affect how it works and how well intended outcomes are achieved. Realist approaches (Pawson and Tilley 1997) highlight the importance of the intervention-context interaction within evaluation, with a range of frameworks operationalizing the domains of context that influence intervention functioning and effectiveness (Pfadenhauer et al. 2017). There is little explicitly realist work on FGCs to date, although some is now emerging (Bernheim et al. 2025; Stabler et al. 2025; Tew et al. 2025). Our work is influenced by this approach, in focusing on the implementation context of FGCs, and especially one key domain, the stage of child welfare involvement. This has some alignment with the epidemiological context domain theorized by the Context and Implementation of Complex Interventions framework, which focuses on the distribution and determinants of needs across populations (Pfadenhauer et al. 2017). As part of this article, we seek to understand if middle-range configurations of context, mechanisms and outcomes work differentially across three stages of intervention, namely (1) early help and child in need; (2) child protection and edge of care; and (3) children in state care. A separate paper (Liao et al. 2026) considers intervention-context interactions for other implementation context domains.

Evidence about FGCs at different stages of child concern

There are several stages of concern about a child where the child welfare system intervenes, moving from lighter-touch early help support, when risk is judged to be relatively low, through to statutory intervention. In the UK, statutory involvement starts with child in need status (or equivalent, as terminology varies across UK nations). When risk is judged to be high, there is more intensive child protection activity. Then, where there is serious actual or potential harm, there is the option of the most serious intervention—a child being moved from their family and into alternative out-of-home state care.

The most well-known child welfare population targeted for FGCs are children at risk of harm. The international FGC evidence syntheses all focus on the evidence for this risk scenario (Dijkstra et al. 2016; McGinn et al. 2020; Nurmatov et al. 2020; Stabler et al. 2025). Wood et al. (2024) found the most common FGC contexts in the UK to be where there was a child protection plan, or care proceedings were initiated or in progress. The largest body of evidence in the above reviews surrounds the prevention of children coming into state care and keeping them with families. Recent UK examples of studies are Mason et al.'s (2017) evaluation of system change in one LA that embraced FGCs for every family who would conventionally have experienced an initial child protection conference, and Taylor et al.'s (2024) randomized controlled trial (RCT) of FGCs at the pre-proceedings stage. This was the first FGC study in the UK using this design and its positive evidence of FGCs reducing care entry has influenced government policy in England (Foundations and What Works Centre for Children and Families 2025).

Other stages of child welfare concern have less research evidence, especially in the UK. Another important stream of FGC work focuses on children in state care preparing to move back to their families—‘reunification’—a scenario featured heavily in the evidence reviews cited above. The Nurmatov et al. (2020) review included some comparative studies from the UK, but these studies were not focused on reunification—all this evidence came from the USA or the Netherlands. However, more than 80 percent of LAs in the UK running FGCs and responding to Wood et al. (2024) survey reported using FGCs for reunification planning.

FGCs are sometimes used for children in care for reasons other than reunification. Lifelong Links is a relatively new UK programme developed by the Family Rights Group, that focuses on support networks for a young person whilst in care and after they have left (Holmes et al. 2020). This may include an FGC with people the child sees as important, whether family or others from their care journey, to plan future support. Evaluations have been positive, in terms of engagement with leaving care support and preventing homelessness and either positive or neutral in terms of placement stability (Holmes et al. 2020; Porter and Fowler 2024; Sanders, Hirneis, and Picker 2024).

Although 71 percent of UK LAs with an FGC service report offering early help FGCs (Wood et al. 2024), there is a dearth of evidence about them at this stage of concern. Internal evaluation of Camden Council’s early help FGCs found that 79 percent of families remained free from further early help or social work intervention in the twelve months following (Early Intervention Foundation 2018). Edwards and Parkinson's (2026) qualitative study of four early help FGCs described positive experiences. As this is a newer development, the same range of evaluations have not yet been applied to early help FGCs as to child protection and edge of care FGCs.

Few studies have explicitly compared FGCs in different child welfare contexts, especially in the UK. An exception is Marsh and Crow’s (1998) study. A subgroup of ‘looked after’ young people (UK term for state care) had more successful FGC outcomes than two different edge of care groups, with the majority of looked after young people having their plans carried out successfully over a 6-month period, with increasing contact leading to reunion with family.

This article asks the question ‘how do family experiences of FGCs and outcomes compare at different stages of child welfare concern?’. There are several reasons to focus on this. Firstly, the use of FGCs is expanding in the UK and there is little evidence to date outside of child protection and the prevention of state care. Secondly, the stage of concern is topical, with the law changing in England to mandate the offer of family group decision-making at one stage only—pre-proceedings, which is a last-chance opportunity to prevent state care. This change is happening in light of positive RCT evidence (Taylor et al. 2024). Thirdly, an assumption emerged from contact with FGC staff during our study that offering FGCs at an earlier stage is likely to be more successful, and it is important to test this assumption.

Methods

The mixed-method study ‘Family Group Conferencing for children and families: Evaluation of implementation, context, and effectiveness (Family VOICE)’ consists of three work packages (WPs) (Scourfield et al. 2024). WP1 involved co-designing of programme theory and evaluation questionnaires in workshops with family members who have had their own FGCs, and with practitioners (Meindl et al., forthcoming). These co-designed questionnaires were then distributed in WP2, the ‘Family Views and Context Study’ (the focus of this article, more on this below). The ‘Service Use Study’ in WP3 is a quasi-experimental comparison of families in LAs that did and did not offer FGCs, looking at outcomes in the children’s social care and education systems two years after. This article does not present any results from the Service Use Study.

Co-design

Workshops were run with family members who had taken part in FGCs in two sites—the London Borough of Camden and Y Bont, a charity in Gwynedd, North West Wales. Workshops with young people who had taken part in FGCs, and with practitioners—FGC coordinators and managers, and local social workers—took place online. In total, fifty-six people participated in workshops: twenty-five adult family members, seven children/young people (aged 10–20), and twenty-four practitioners. Each group attended two workshops. The first was focused on developing a logic model for FGCs and the second on prioritizing questionnaire measures. The latter included discussion and ranking of some standardized measures.

Family views and context study

Eighteen LAs from all four UK nations agreed to take part in this study. FGC coordinators invited families referred for FGCs to complete questionnaires. Questionnaires were distributed at three time points: the weeks leading up to the FGC (T1), two weeks after the FGC (T2) and six months after the FGC (T3). There were thirteen different versions of the questionnaire for different ages and time points. Almost all were completed online (99.7 percent at T1 and 99.6 percent at T2) but a shorter telephone version was available. The FGC coordinators also completed a coordinator survey at T2. Families were successfully recruited in fifteen of the eighteen LAs. Based on estimates from most of the participating LAs of the number of families invited by FGC practitioners to take part in the study, the family response rate was approximately 26 percent. Two hundred and sixteen families completed T1 questionnaires, but only 155 of these (380 individuals) went on to have an FGC and consented to participation in the study. Some families’ data could not be included because of missing FGC coordinator data. At T2, 119 families (254 individuals) completed questionnaires, and ninety-nine families (168 individuals) did so at T3 (Supplementary Material Figure 1, see online supplementary data). Recognizing the importance of the wider family and friends network to the FGC process, anyone involved in the FGC could complete a questionnaire, but questions were tailored appropriately—for example, some items only appeared for respondents who said they were living with the child(ren). Each time respondents completed a survey, they received a £10 shopping voucher. Ethical approval was granted by the UK Health Research Authority’s Social Care Ethics Committee (REC reference 23/IEC08/0029).

Questionnaire variables

Supplementary Material Table 1 (see online supplementary data) lists all questionnaire variables, time points (T1, T2, and/or T3), sources (family and/or FGC coordinator), how they were measured, and how they were used in the statistical analysis. Some of the actual data collected were more fine-grained, but needed to be re-coded into fewer categories so that sample sizes were large enough for meaningful analysis. The carer self-efficacy scale only appeared for adults living with the child(ren), and child psycho-social quality of life appeared for children and young people and for adults living with them.

Qualitative data collection

In the T2 questionnaire, family members who had recently participated in an FGC were asked if they would participate in an additional interview. In total, 186 expressed an initial interest and all were approached by email. A purposive sampling approach was adopted, aiming to recruit roughly equal numbers of people (around ten) from each of three broad categories in terms of the stage of child welfare intervention, as noted in Supplementary Material Table 1 (see online supplementary data): (1) early help and child in need (EH/CIN); (2) child protection and edge of care (CP/EC); and (3) child looked after (CLA). The target of around thirty was pragmatic and based on previous experience of sample size needed to cover a range of experiences and likely to achieve saturation. Each interview participant was offered a £20 voucher as an incentive and ‘thank you’.

Over a 3-month period, thirty-one semi-structured individual telephone or online interviews were conducted. A further four individuals came forward who were not required due to saturation having been achieved. Although any FGC participant, regardless of their relationship to the FGC child, was eligible, the original aspiration had been to cover a range of genders and family relationships. However, insufficient numbers came forward to make selecting possible, so the final achieved sample had an element of convenience in relation to factors other than the stage of child welfare process. Interviewees were recruited from twelve of the fifteen participating LAs and from each of the four UK nations. Written consent could not be obtained from six of the thirty-one participants, so these interviews were deleted, leaving twenty-five interview participants whose demographic characteristics are noted in Supplementary Material Table 2 (see online supplementary data).

Interviewees were asked to share their experiences of the FGC meeting and the role of the coordinator, as well as any marked changes that had taken place in their lives since the FGC. The structure of the interview schedule (see https://doi.org/10.17035/cardiff.32257299) informed the approach to the analysis of the data (see below), and the presentation of the findings. Topics covered included the coordinator’s role, how the FGC meeting went, the role of the local authority, reflections on what has happened since the FGC and views on how the FGC experience could be improved.

Analysis

Quantitative analysis of questionnaire data was conducted in two stages. First, aspects of the FGC context and outcomes were categorized according to the stage of concern about the child(ren), and descriptive statistics were produced. Second, statistical testing was performed. A Shapiro–Wilk test was first used to examine whether continuous variables were normally distributed. The results indicated significant deviations from normality (P < .05) for at least one of the three stages of welfare concern for each dependent variable. Accordingly, Kruskal–Wallis tests were used to assess differences between the three stages. Where statistically significant results were found (P < .05), Dunn’s post-hoc tests with Bonferroni correction were conducted to identify which categories differed. Analyses were carried out using Python software.

For proportions derived from count data, chi-square tests were used to examine differences across the three categories of the child welfare process. Where the overall chi-square test indicated significant differences between the three categories, pairwise chi-square tests were subsequently performed between each pair of categories to determine where these differences occurred. The chi-square test was appropriate as all variables had total sample sizes ≥20 and met the assumption that all expected cell counts were ≥5 (or at least 80 percent were ≥5 with none <1).

Outcomes measured at all three time points were child psychosocial health, adult wellbeing, self-efficacy, and family functioning. Change was calculated from T1 to T2 and from T1 to T3. The analysis was at a family level rather than individual level, so where multiple individuals from the same family completed questionnaires, the mean response per family was used. For the child psychosocial health scores, a combined score was calculated using responses from both children and adults within the same family. When both sets of responses were available, each (children and adults) was weighted equally at 50 percent.

The semi-structured interviews were transcribed, and anonymized transcripts imported to QSR NVivo V.12 software to support coding. Thematic qualitative analysis was adopted, focusing upon identifying, describing, and analysing both implicit and explicit themes within the data set (Guest, MacQueen, and Namey 2012). This involved two processes; firstly, descriptive themes were identified and created from participants’ experiences of the FGC meeting and the coordinator’s role. Secondly, there was interpretation of reported changes since the FGC, mapping these to mechanisms and outcomes identified in the study’s initial logic model, whilst at the same time allowing for the emergence of new mechanisms and outcomes from the data and being open to unanticipated themes. In both processes, analysis involved identifying both similarities and differences across the three stages of child welfare concern.

Quantitative and qualitative data were analysed simultaneously, then results brought together and compared in an integrated interpretation process. This would generally be regarded as a convergent parallel mixed-methods process (Creswell and Plano Clark 2018).

Findings

Quantitative research

Of the 155 families providing data for analysis, most were involved in either the early help/child in need (n = 57) or child protection/edge of care stages (n = 64), with smaller numbers of looked after child FGCs (n = 31) (Supplementary Material Figure 1, see online supplementary data). Table 1 shows variation both in FGC characteristics and outcomes by stage of concern and Table 2 shows results for the subset of outcomes that were measured at all three time points.

The most striking and consistent finding was a lack of significant difference between the three stages of concern, on almost any measure. Table 1 shows that across all three categories, families consistently reported that they had a say in FGC venue, medium (online or in-person) and who got invited to the meetings. Numbers attending from different branches of the family were broadly similar across the three stages. Rating of communication, both within the family and between the family and social workers, was similar. Social workers’ understanding of the family (from family members' perspective) was lowest for the child protection/edge of care category, as might be expected, but differences between the three stages in response to this item were not statistically significant. Whether the plan was implemented, the child was thought to be safe and the FGC was thought to have a role in this were also similar across the three stages of concern. Overall satisfaction was consistently high, with mean ratings of four on a five-point scale.

In general, in all three categories, there were only very small and statistically insignificant changes over time in the outcomes that were measured at multiple time points (Table 2). Exceptions include changes in self-efficacy between T1 and T3. Dunn’s test (Supplementary Material Table 3, see online supplementary data) showed a significant decrease in self-efficacy between T1 and T3 for the early help group compared to the child looked after group (P = .039). It is important to note, however, that the sample size for the child looked after group was very small at T3 (n = 8).

Qualitative research

Themes from the qualitative analysis explored participants’ experiences of the FGC process, central mechanisms, and intervention outcomes. All three sets of themes focus on similarities and differences across the three stages of child welfare concern: (1) early help and child in need (EH/CIN); (2) child protection and edge of care (CP/EC); (3) child looked after (CLA).

Experiences of the FGC

Overall participants reported positive experiences of FGCs, with notable areas of positive experience in three areas of participation.

Participants across all stages of concern reported a positive experience of the preparation for the FGC meeting by the coordinator and practical arrangements. Positive aspects of preparation found across all three categories included timing and the forms and level of communication, EH/CIN 5/8 people, CP/EC 6/9 people, CLA 5/8 people. An example was this mother:

She did a great job… She made sure that we were in constant communication. So, we’ve spoken via WhatsApp, via telephone, via email. But she always makes sure to give us plenty of notice for things. (Mother 11, CP/EC)

The only negative experience reported involved a mother from the CLA category who experienced a delay in the organizing of the FGC:

I think it was eight weeks or six weeks, so, it was constantly in my head, and I feel like we could have worked it out as a family. (Mother 17, CLA)

Within this period of communication between coordinator and family members, a consistent finding across all three categories was the level of flexibility afforded, in relation to the timing and format of the meeting:

He worked around like my schedule and what I could do and what I couldn’t do. (Mother 1, EH/CIN)

For several families, a flexible format was important in considering geography, as well as personal and professional commitments, as reflected in the following extract:

She prepared and sorted out when we could all get there… it was somewhere far from where we live. She sorted all that out, also arranged a Zoom call for my one daughter that couldn’t get there. The other one responded by a letter, because she’s a nurse and didn’t think she’d be able to get out. (Grandparent 25, CLA)

Similarly, where there were issues of the safety or the best interest of family members, separate meetings were sometimes used:

That was another thing that I did appreciate, the fact that they took the time to do the conferences separately… because I would not sit down in the same room as him… (Mother 6, CLA)

Participants indicated a positive experience related to the role of the coordinator in the FGC meeting itself. This was consistently reported across all three stages of concern, with no evidence of negative cases. The role of the coordinator was seen as crucial in ensuring the democratic nature of the conference, where everyone was given equal opportunity to express their feelings and opinions. For the following mother, it was particularly important that this process involved everyone having the opportunity to ask any questions:

She made sure that everyone got their point across and what needed to be done going forward… like everyone that had a question managed to ask the questions and get an answer back. (Mother 1, EH/CIN)

Also found across all three categories was the importance of the coordinator’s skill in achieving a balance of involvement, as expressed in the following extract:

He did get involved, but not in an invasive way. He sort of like helped us see the other side of things in a sort of thing. So, he did get involved but not never felt like he was stepping over the line, or… or anything like that. (Mother 6, CLA)

Crucial to achieving this balance was allowing family members the physical time and space to discuss, and find their own solutions, with the coordinator leaving the meeting for a period of time. This is what is termed ‘private family time’:

Like, I know that within the meeting, but she didn’t rush us with it, she didn’t say right, you’ve got half an hour. She’d say take as long as you need and just give us a call when you would like me to come back in. (Aunt 13, EH/CIN)

Participants across the three stages reflected positively on how the FGC offered a safe environment to talk (EH/CIN 6/8 people, CP/EC 6/9 people, CLA 5/8 people), with no negative experiences reported. A notion of safety was reported in the meetings, where people were able to express their emotions and feelings in the presence of family members and professionals:

I was able to open up and speak about the situation and how I was feeling. Yeah, it’s definitely made it a safe environment now to be able to speak about everything that’s gone on. (Mother 2, EH/CIN)

Similarly, the FGC meeting provided a space to share difficult experiences with other family members who had not been previously aware of these difficulties, as in the following extract:

I don’t know how much they knew about what was going on, but I suppose it’s a space where they can… you know, they… they learn what you’re going through, don’t they? (Mother 20, CP/EC)

Perceived intervention mechanisms

Participants indicated two central mechanisms through which they thought the intervention would generate outcomes.

The first mechanism identified, predominantly found in the interviews with family members in the EH/CIN and CLA stages, involved experiencing a sense of inclusion and having a ‘voice’ in the family, following the FGC meeting. The activation of this mechanism appears to be the safe environment the FGC meeting created that allowed experiences and feelings to be both shared and heard, for example, in the following extracts:

I feel more included with the situation, as whereas before I was kind of like, an outcast, but after the meeting, I got given the role to kind of keep grandad afloat and attend meetings with him, so by doing conference, it gave me that opportunity to say I’d like to be there for grandfather. (Aunt 13 EH/CIN)

I feel really listened to. I feel supported… I think what helped with this family group conference… because with this situation, my child was coming home to me, so I did feel listened to, because ultimately, I’m in the driving seat. (Mother 28, CLA)

Similarly, for the following mother the meeting offered an opportunity to re-address some of the power imbalance she felt between herself and social workers and provide her with a ‘voice’:

I didn’t really like confrontation with speaking to social workers… it was hard for me to, like, speak up and… and be heard. I just used to get on with it. But it’s made it totally different where I’ve got a voice and I can speak and I’m the parental guardian, and yes, some things I might say you might not agree with but I’m still… I’m still… I’m still going to voice my opinion. (Mother 15, EH/CIN)

The second mechanism concerned how the FGC provided family members and friends with an opportunity to consider their motivation and ability to offer support to the parent/s and child who was the focus of the FGC. This mechanism emerged predominantly from the statutory contexts of CP/EC and CLA:

It was quite interesting, like, I think… because we all know each other anyway, but it was quite interesting for us all to be sat in a room and kind of understand what we’re all offering to help each other, and probably all showed how each other’s feeling about the situation. (Family friend 7, CLA)

For the following aunt, an activation of this mechanism resulted in her deciding not to support her sister again, through her reflecting upon a previous FGC meeting and the support she provided:

I think it was a good springboard, and, you know, if it’s your first time around the block, absolutely a great place, but not for round two. It was no reflection of the coordinator or the process, but I’m absolutely having nothing to do with it now… round two for me, and I can’t do it again… (Aunt 9, CP/EC)

Perceived intervention outcomes

Participants reported two key outcomes activated by the perceived mechanisms.

Participants, especially in the EH/CIN category (5/8 people), described increased communication between family members, and the re-building of previously distant or troubled relationships. For example, in the following extract a mother reflected upon her improved relationship with her father:

It’s also built a relationship between me and my dad, because we didn’t have the best relationship as I was growing up. So, it’s brought our relationship together as well. So, being able to speak about and sort of opening up, it sort of like built a bridge. (Mother 1, EH/CIN)

The following stepfather reflected upon how the meeting allowed him to be open and honest about the problems he was facing permitted a more positive relationship with other family members:

At first, we were lying about the drugs and stuff, and then once we come out clean, now it seems everything is going along quite smoothly. (Stepfather 10, EH/CIN)

Similarly, a further outcome was increased support from family members and friends following the FGC.

Participants across all three stages reported increased family support. (EH/CIN 4/8 people, CP/EC 4/9 people, CLA 4/8 people). An example is in the following extract from a mother who reflected upon increased support from her parents:

They have them regular, so if I just say to them, ‘Oh, can you have them for this? I…’ or if I ring them up and say, ‘Look, can you have them for a couple of hours?’ Before the conference, it would have been a no… So, it has changed big time. (Mother 2, EH/CIN)

However, a number of family members with the EH/CIN stage (4/8 people) reported a lack of support from their LA following the FGC, with family members experiencing an over-reliance on them, as experienced by this aunt, who suggested that the FGC was the platform by which this occurs:

I don’t know, for me, it was… it was social services handed off the job to someone else again, that’s what those things are designed to do, and… yeah, sometimes that’s not the help that’s needed. (Aunt 13, EH/CIN)

The other negative experiences reported in interviews did not actually concern the FGC per se; rather they related to the LA social workers involved with the families. Predominantly found in the contexts of CP/EC (6/9 people) and CLA (5/8 people), this negativity involved family members not wanting social workers in the FGC meeting, because of not trusting them; or social workers not implementing the care plan following the FGC; or annoyance at the number of changes in social workers both during and after the FGC process.

Discussion

The Family Views and Context study considered how implementation context affects FGC experiences and outcomes, with this article focusing on the specific context of the stage of child welfare concern. To our knowledge, the study had the largest sample of families reporting their experiences of any UK FGC study to date, although others (e.g. Mason et al. 2017; Taylor et al. 2024) have had larger samples from administrative data.

The overall finding is of little variation across the three stages of: early help and child in need; child protection and edge of care; and child looked after. This was the case for quantitative evidence from questionnaires about implementation factors and outcomes, and also qualitative interviews, where families across stages described broadly similar positive experiences of preparation, meeting conduct, and a sense of voice within a safe forum. Taken together, the pattern points to common mechanisms that may operate irrespective of stage.

Two configurations of context, mechanisms, and outcomes, consistent with our logic model can be tentatively identified from the qualitative interviews. Firstly, when coordinators prepare for FGCs early and flexibly (C) a sense of ownership is fostered amongst wider family members (M), which in turn supports within families clearer communication, broader involvement in care for children, and pragmatic planning (O). Secondly, the intervention components of a structured meeting, balanced facilitation, and private family time, successfully delivered through skilful and proactive coordinator practice (C) enabled psychological safety and voice (M), producing commitments from wider family members to concrete support and improved intra-family communication (O). The core mechanisms were fairly consistently reported in interviews and supported by quantitative analysis, suggesting that how FGCs are delivered may matter more than when they are offered along the pathway of social work involvement—at least for the proximal outcomes measured here. Both CMOs are in line with Stabler et al.'s (2025) realist review of evidence on how FGCs can prevent out-of-home care, so we theorize FGCs can operate in a similar way at other stages of concern.

To return to the overall finding about little difference between the categories, statistical tests found no significant differences for seven out of eight of the outcomes measured at three time points. The one significant difference was also based on only eight families in the child looked after category at the six-month follow up, so should not be taken as definitive evidence. Qualitative interviews consistently found positive experiences across all categories of the preparation for and conduct of the FGC meeting, as well as the meeting being a safe place for opinions to be heard. Changes in outcomes measured at multiple time points were small; however, high ratings were given for all three categories for perceived child safety (and the FGC’s role in that) and for families having a say, both in practical FGC arrangements and at the meeting itself. Family participation in important decisions about children is seen by many people, including families themselves (Mitchell 2020) as the main purpose of an FGC and an important outcome in its own right.

The early help category is where there has been the least research evidence internationally to date. Our study provides some preliminary evidence that FGCs may be associated with some positive outcomes for these families, as seen in questionnaire results about satisfaction, participation, communication, child psychosocial health, and short-term gain in self-efficacy; as well as qualitative findings about improved communication and support within the family. The positive qualitative findings echoed those of many other studies over the years about FGCs at later stages of child welfare concern (e.g. Holland et al. 2005; Mutter et al. 2008; Mason et al. 2017; Stabler et al. 2025).

In summary, our findings point to common, delivery-dependent mechanisms—notably high-quality preparation, balanced facilitation, and private family time—that appear to operate across different stages of involvement. Within the limits of statistical power, measurement, and category structure, we found very few stage-based differences on the proximal outcomes assessed.

Limitations

There are several important limitations to note. First, the response rate to the quantitative survey was low. Declining response rates are unfortunately a global trend, explained at least in part by the proliferation of online surveys (Eggleston 2024). On reflection, if the research team had worked more intensively with fewer LAs, the response rate might have improved. Second, the Family Views and Context study is a single-arm study with no comparison group of families who did not receive FGCs. Third, the grouping we decided on for the stages of child welfare concern will have an important impact on the results. If we had divided up the sample differently, we might have seen more variation between categories. Within our categories there was a large range of experience—for example, the early help and child in need category included some families with relatively low-level problems and others with much more serious problems, where a child protection plan seemed likely. However, on balance, our groupings have some credibility and did mostly provide sufficient sample sizes for comparative analysis.

A fourth issue was potential selection bias in the interview sample. Around one in six of those who had indicated their willingness to be interviewed then responded about interview arrangements. These were very positive about their FGC experience, but it could well be that those with more positive experiences were disproportionately willing to be interviewed. Finally, we note that family members from the early help/child in need category were more vocal and engaged with the interview, whereas family members from the child protection/edge of care category appeared more reluctant to share in detail their experiences of the FGC process. Studies into parents’ views of the child protection process have found them experiencing the system as uncaring and inflexible, with a power imbalance that often left them with a sense of shame and a reluctance to speak out, or talk about their experiences to professionals and family members (Smithson and Gibson 2017; Gibson 2020). It could be that research interviews including questions about the child protection process evoked feelings of shame and therefore a reluctance to share.

Implications for policy and practice

These findings could be taken as broadly positive by proponents of FGCs. Whereas the Government in England is about to mandate FGCs only at the pre-proceedings stage, our overall findings of some similar impacts of FGCs across all stages of the child welfare process could argue for more offers of FGCs to a greater number of families, beyond the fairly narrow context of pre-proceedings. This would increase family voice in a child welfare system where decisions about children tend to be professionally led. In particular, the early help/child in need group was previously the one least researched, and these families showed some promising outcomes, although as with the other child welfare categories, the outcomes measured at three time points showed relatively little change. FGCs are only a relatively light-touch intervention and separate analysis from the same study (Liao et al. 2026) shows the importance for positive family outcomes of other support services in addition to FGCs.

At least until we have more evidence about these questions, LAs and policymakers should therefore prioritize consistent delivery standards and consider an offer of FGCs across the stages, while remaining alert to stage-specific needs (for example, relationship repair in a child protection context; expectation management for children looked after). There is a risk that without additional resources, offering more FGCs may dilute the very qualities that make them effective for families at the child protection/edge of care stage. If full-quality widespread provision is not possible, given resource constraints, then our findings could also be interpreted as an argument for focusing on quality over quantity.

Implications for research

More in-depth research would be helpful to further compare FGCs at these different stages of child welfare concern. This could include use of qualitative methods such as observation, and interviews following families over time; and quantitative analysis of administrative data would provide larger datasets and better access to data on system outcomes. Our forthcoming Service Use Study (Scourfield et al. 2024) use administrative data to measure service outcomes of FGCs at various stages of child concern, but more is needed.