Exploring the declining rates of state social work intervention in an English local authority using Family Group Conferences

A Over the last decade England has seen rising numbers and rates of children in care and subject to child protection intervention. This article draws on national data and quantitative data from a study of ‘City’, a local authority in the north of England with a large Family Group Conference service. In distinction to national trends, City has seen substantial declines in the numbers and rates of children in care and who are subject to child protection intervention. The article analyses how notable these declines may be, comparing the declines with national trends, and exploring six-month service indicators for children who had a Family Group Conference. Implications regarding the reduction of levels of state social work intervention in family life are discussed.


Introduction
This article draws on analysis of public UK Government data on child and family social work intervention rates in England alongside quantitative data gathered as part of an evaluation of 'City', an urban local authority in the north of England with a large Family Group Conference (FGC) service. City used FGCs as part of an approach called Restorative Practice which can be summarised as a 'high challenge/ high support' way of working (Wachtel & McCold, 2001). Applied to practice with families in City the focus of this approach was on providing families with a relational way of working where child welfare concerns were raised with families, but they were given the support to try to arrive at their own solutions to them. As part of this approach City had explicitly articulated the aim of reducing the numbers of children living in state care in its area. Growing concern about the rising numbers and proportions of children in state care in England have been reflected in a recent review of the child care system (Care Crisis Review, 2018). Such concern is situated amongst a wider set of questions internationally regarding child protection systems which have become overly driven by an investigative focus and which give families too little support to address identified concerns (Bilson & Hunter Munro, 2018). While this issue has had recent attention it is part of a longer standing concern in England: the Children Act 1989 and the 're-focussing initiative' in the mid-1990s were both introduced against a backdrop of questions about narrowly focussed child protection practice. Recent concern has been generated by a number of factors including: the proportions of children in state care rising by over a third In the UK FGCs have been used since the 1990s, but while the FGC process is heavily influenced by that used in New Zealand, it differs in that they do not have a statutory basis. As a result it is down to individual local authorities to decide if they should offer FGCs to families, and if so in what circumstances. This has resulted in substantial variation in their use according to locality. The offer of an FGC is also optional for families in the UK -they can choose whether to have one or not and pre-existing national regulations about child protection processes must apply whether or not a family accepts the offer of an FGC. City operated a 'hybrid system' of FGC use, by which is meant two things: firstly that it was mandatory for social workers to refer families for an FGC at certain points in their involvement, such as when an Initial Child Protection Conference (ICPC) had been called, but it was optional for families as to whether they accepted the referral and participated in the FGC, unlike in New Zealand. Secondly that, regardless of whether families had an FGC, as noted above, they were still subject to standard child protection processes governed by national guidelines. This article draws on data from the first significant study of FGC use in City, the largest ever local authority FGC service in the UK. The developments are important in terms of their scale and the findings significant in considering the potential impact that the implementation of participative decision-making fora may have when operating within a hybrid system and a wider context of rising social work intervention levels.

Care numbers and social work intervention levels in England
There are three levels of statutory child and family social work intervention in England. The lowest level occurs when a child is classed as a 'Child in Need' (CIN) and statutory social work involvement is provided on the basis it is required to support a child's health and development, or because a child has a disability. The next level is where a child becomes the subject of a Child Protection Plan (CP) (formerly known as the Child Protection Register) following a multi-agency Initial Child Protection Conference (ICPC) on the grounds that the child is suffering, or likely to suffer, significant harm. An ICPC should, under national guidelines, be held within 15 working days where it is thought a child is likely to suffer significant harm (HM Government, 2018). This level of intervention typically signifies that a child is living in parental care with the highest level of concern about that arrangement without it meeting the threshold to place a child in state care. Finally, a child may be placed in state care via foster care or residential care whereby they become a Looked after Child (LAC). Though there may be a small number of situations where a child has both CP and LAC status for short periods of time, children will normally only be in one of the three categories at any one time. LAC, CP and CIN rates are all given as the number of children in a geographical area per 10,000 of the child population in that area.
Though there are no explicit legal duties on local authorities in England to prevent children entering state care, the core principles of working in partnership with parents and minimal state intervention within the Children Act 1989 imply the desirability of maintaining children in parental care where this is consistent with their welfare. In the years following the introduction of the 1989 Act, the LAC rate in England fell from 60 per 10,000 children in the late-1980s to a rate in the 40s in the mid/late 1990s, before it started rising again (Rowlands & Statham, 2009;Department of Education (DfE), 2018). Since the late 2000s, the LAC rate has risen consistently, if steadily, from the mid-50s, to a rate of 64 in the latest figures for 2018 (Rowlands & Statham, 2009;Department of Education (DfE), 2018), the highest rate for over 30 years. Analysis has shown that there was a statistically significant increase in the proportions of children entering state care in England between 1992 and 2012, combined with children spending longer periods in care than in the 1990s (McGrath-Lone, Dearden, Nasim, Harron, & Gilbert, 2016). Increasing numbers of children have also left the care system in England for legally permanent arrangements outside parental care: for example between 2013 and 2018 over 20% of the children who left care in England did so to enter Special Guardianship arrangements or adoptive placements (Department of Education (DfE), 2018). Therefore, the rising LAC rates represented in government statistics only partially reflect the increasing numbers of children in England who are living permanently outside parental care following state intervention due to child welfare concerns (Bilson & Hunter Munro, 2018).
Rising LAC rates in England have also occurred alongside rising CP rates and increases in child protection referrals. A recent national survey of local authorities in England found that from 2008 to 18 the numbers of referrals to children's social care had increased by 22%, the number of Section 47 (child protection) inquiries increased by 159% and the number of children subject to CP intervention by 87% (Association of Directors of Children's Services, 2018). While some of these rises may be due to demographic changes such as a rising child population in the UK from 2007 (Office for National Statistics, 2019) they do not account for this level of increased child protection activity.
It is worth nothing that increases in the number of children referred to child protection services in England are mirrored in a number of other developed countries in the 2000s (Bilson & Hunter Munro, 2018), and there has also been a general trend towards rising proportions of children living in state care since the late 2000s in western European countries (Burns, Pösö, & Skivenes, 2017). However, rising rates of state care usage are not uniform across all developed countries: Denmark (Ubbesen, Gilbert, & Thoburn, 2015) and the USA (U.S. Department of Health & Human Services, 2018), two countries with highly different regimes of welfare support, have not recorded sustained increases in numbers in care since the late 2000s. While there does appear to be a trend towards greater rates of children in care, therefore, there is also some divergence which highlights the need to more closely consider contextual factors within countries which may have contributed to rising rates. Most obviously in England austerity has seen cuts to many poorer families' incomes, while unprecedented government cuts to local authority funding have resulted in drastic reductions to services supporting families. This is illustrated by recent fiscal analysis of government spending from 2009 to 10 which reported that funding for Children's Services in England per child is due to reduce in real terms by 20% by 2019-20, while funding for Early Intervention and preventative services had already fallen by around 60% in real terms by 2016-17 (Kelly, Lee, Sibieta, & Waters, 2018). Such decreased funding has occurred at the same time as increasing referrals to Children's Services and rising social work intervention levels (Bilson & Hunter Munro, 2018). As a result there are systemic pressures at all levels of the English child protection system, leading to concerns about how it can cope with the demand, alongside questions about the ethical basis for such increased levels of statutory intervention and surveillance in families' lives.

FGCs and the contested evidence base
The current international evidence suggests positive family and professional experience of FGCs as a process but mixed evidence of their immediate and intermediate impact on child welfare outcomes. There is presently a range of empirical evidence on FGCs with over seventy peer reviewed empirical studies published in English at the current time. A majority of these employ quantitative methods. However, only a minority are outcome studies with data on traditionally measured child welfare outcomes such as child maltreatment rates, rates of entry into state care, children's placement pathways and the level of involvement of state social work agencies. In respect of the UK, there is a predominance of qualitative studies and the only published work containing some, proxy, indicators of child welfare outcomes connected to FGCs are Marsh and Crow's (1998) mixed methods study of early FGC pilot projects in England, the current study (Mason, Ferguson, Morris, Munton, & Sen, 2017) and a further Government study of FGCs elsewhere in England commissioned as part of the same government initiative which funded the current study (Munro, Meetoo, Quy, & Simon, 2017). The potential of FGCs to positively impact on child welfare outcomes was suggested by Marsh and Crow's (1998) early summary of pilot FGCs projects as well as Pennell and Burford's (2000) pioneering development of them in Canada. Both found FGC use was linked to reduced child maltreatment rates, lower numbers of children entering state care and lower levels of state social work involvement, where it continued. However, two later studies in Sweden and the USA, with randomised comparison groups, found neutral effects in respect of the length of time children were in out-of-home care, rates of child maltreatment and the amount of state social work involvement (Berzin, 2006;Sundell & Vinnerljung, 2004). These more pessimistic findings were reinforced by the only current meta-analysis of FGC outcomes (Dijkstra, Creemers, Asscher, Deković, & Stams, 2016: 108) in which the authors conclude with the grand claim that 'Robust research into the effectiveness of FGC is so limited that it does not support its widespread use.' Such a claim is heavily contested. De Jong & Schout, 2018 question whether Randomised Control Trials (RCTs), often claimed as the gold standard of effectiveness trials, are an appropriate method for capturing the operation of a complex process like an FGC. They argue that FGCs operate in open systems, where outcomes patterns are liable to be affected by multiple interacting factors, which cannot all be controlled for. They propose that a more qualitatively influenced paradigm, which appreciates the different 'lifeworlds' of the various actors in the FGC process, is better suited to capturing its complexity, and argue that such a paradigm entails acceptance that there can be multiple, divergent, experiences and views of the same phenomenon.
Research which has examined the FGC process illustrates some of the factors which may affect whether positive outcome patterns subsequently develop, including the preparation of family members for the FGC, whether there is sufficient support for children to participate fully, the independence of FGC co-ordinators from state social work provision and the approval of, and support for, family plans after an FGC (Connolly, 2006;Dalrymple, 2002;Pennell & Burford, 2000). Findings suggest most families who have had an FGC feel they offer greater opportunity for their voices to be heard and to arrive at family-driven solutions (e.g. Holland, Scourfield, O'Neill, & Pithouse, 2005;Mason et al., 2017;Pennell & Burford, 2000). There is also some evidence FGCs can lead to greater social support (Corwin et al., 2019). Family meetings are typically attended by a wider circle of family and friends than routine child protection meetings and can effectively include children and young people in ways they find hugely important (Dalrymple, 2002;Holland & O'Neill, 2006). Empirical evidence also illustrates how families may judge the success of FGCs by more personal criteria than professionals and researchers, such as whether they feel empowered within the emotional dynamics of a meeting, whether the FGC holds particular family members to account, the impact of the meeting on family relationships and whether more practical actions and arrangements occur following the meeting (Holland & O'Neill, 2006;Mitchell, 2019).
The relatively limited evidence on children's experience of FGCs also illustrates that different family members can have notably differential experiences of a meeting: some children have reported feelings of distress after observing adult conflict in FGCs, a reticence to express views at the meetings, and a sense that they were more included than actively influencing decisions (Dalrymple, 2002;Holland et al., 2005; Merkel-Holguin, Schwab-Reese, Drury, Allan, & Hollinshead, 2019). The need to continue to give close attention to the complex dynamics of children's empowerment within a participative family decision making forum is thereby highlighted, with some proposing that the use of independent child advocates could be required within the FGC process to support children to have their perspectives fully reflected within family plans (Dalrymple, 2002;Holland et al., 2005). Overall, the research evidence suggests the FGC as a multi-faceted encounter, valued by most families who take part in one. While they do not always resolve family difficulties they do tend to facilitate family members to come together to discuss concerns in a way that can widen networks of support and monitoring and lead to family-generated proposals which address the child welfare concerns raised by state agencies (Corwin et al., 2019). This contrasts markedly with parents' almost universally negative reported experiences of routine child protection processes and meetings (e.g. Broadhurst, Holt, & Doherty, 2012;(Sen, 2016)).
It has also been proposed that instead of viewing FGCs as an 'intervention' to be evaluated by effectiveness studies they should be seen as a process which is part of a family's civic rights in a liberal democracy where there is compulsory state intervention on child welfare grounds (Morris & Connolly, 2012). The establishment of FGCs in New Zealand was premised on their qualities as an ethical, rather than effective, practice (Morris & Connolly, 2012). The Children's and Young People's Well-being Act 1989, which introduced FGCs in New Zealand, framed them as a rights-based process within the post-colonial settlement there, in a context where Maori children were heavily over-represented in the care system (Barrington, 2004). It was argued they recognised a family's rights to generate family solutions to the concerns raised by the state, and recognised children and families' rights to preserve kinship links (Barrington, 2004). Judy Keally, the Chairperson of the select committee overseeing the 1989 New Zealand Bill, commented that FGCs would give families an opportunity to defend themselves against 'over-zealous professional intervention, and give them a chance to solve problems themselves' (cited in Barrington, 2004, p.16). One premise underpinning this article is that viewing FGCs as a rights based process should not entail ignoring considerations of the outcome patterns associated with their use. There was more than a suggestion in the development of the 1989 New Zealand Act that if families were given the opportunity to come together to design family solutions to professional concerns then at least some would respond positively to it (Barrington, 2004). Instead, our argument is that data on outcome patterns associated with FGCs can be useful for understanding how and why they may operate in particular ways in given contexts (De Jong & Schout, 2018;Pawson & Tilley, 1997). However, such data should be used to develop greater insight into how families can be better supported to arrive at their own solutions to state-mandated child welfare concerns, not to question families' fundamental rights to be so supported.

The study
The study on which this article is based was a wider Government funded evaluation of an Innovations Programme investment in City from 2015 to 16 to extend the use of Restorative Practice within Children's Services there, including via an expansion of the FGC service (Mason et al., 2017). The FGC strand of the evaluation was mixed methods and explored the impact of upscaling the use of FGCs in City by creating an additional fourth FGC team. Ethical approval was granted via City's ethical governance arrangements and through the University of Nottingham. Data collection for the FGC strand of the evaluation included: • An analysis of administrative data held by the FGC service for all children and families referred to the service in the 2014 and 2015 years; • 15 days of practice observation in the FGC teams over three months; • Informal discussions with co-ordinators about their work pre and post observation of direct practice; • The development of ten case studies of families who were tracked over the observation period; • Structured repeat focus groups with co-ordinators convened at three points over the study; R. Sen and C. Webb Children and Youth Services Review 106 (2019) 104458 3 • Semi-structured interviews (n = 39) and questionnaires (n = 66) with FGC co-ordinators and managers; • Structured telephone interviews with adult family members who had previously worked with the FGC service on their experiences (n = 36).
Data from the study regarding the use of FGCs in situations of family violence have previously been published (Sen, Morris, Burford, Featherstone, & Webb, 2018). This article draws on quantitative data from the study derived from analysis of the administrative data held by the FGC service in City regarding the initial and six-month service indicators for children who had an FGC. Data from the FGC service database for the 2014 and 2015 years were combined in Excel and used to generate descriptive statistics on the required indicators. The database included information on 1267 children who progressed to an FGC in City in 2014 and 2015. In addition, this article firstly presents comparative analysis of changes in overall social work intervention rates in City and national trends, based on public data from Looked after and Child Protection censuses in England on LAC, CP and CIN numbers and rates (Department for Education (DfE), 2010a, 2010b; Department for Education (DfE), 2015a, 2015b; Department of Education (DfE), 2018; Department of Education (DfE), 2019). This analysis consists of modelling changes in LAC rates over time between 2010 and 2015 using a latent growth curve model with random intercepts, slopes, and quadratic terms for each local authority (Little, 2013), before extracting factor scores and comparing local authorities in order to contextualise City's trends.

The model of FGC use in City
The FGC service in City primarily received referrals from the child and family social work teams. After referral, an FGC co-ordinator met with the social worker to discuss the key concerns which underlay social work involvement and then contacted the primary carer(s) for the child to see whether they would meet with the co-ordinator to discuss the possibility of an FGC meeting being convened. If the carer agreed, the referral then became an open case to the FGC service and was then deemed an 'FGC enquiry'. If the primary carer(s) then agreed to proceed to a full FGC meeting there was discussion, and sometimes negotiation, about which other adults within the family network should be invited, meetings with those adults to discuss the issues behind the FGC, and discussions regarding the best ways of involving the children in the meeting.
FGC co-ordinators worked as part of Children's Services in City but in discrete teams from social workers. Their independence from social work services was underpinned by the fact that discussions with families regarding an FGC were not recorded or shared with social workers so that families were able to consider participation in an FGC without those deliberations potentially becoming part of later court or social work processes. The FGC service in City first became a small local authority-wide service in 2010. It expanded in 2014 to comprise three separate FGC teams each located in the three core geographical areas of City, and was further expanded to a fourth team in late 2015 through the Government Innovation Fund described above. The fourth team had a remit to focus on families where domestic violence was an issue and to start to take referrals from 'Early Help' rather than social work sources. The expansion of the FGC service from 2010 was a central part of the adoption of Restorative Practice in City. Training on Restorative Practice had started to be rolled out in 2011, and its use became widespread from 2012. While the use of FGCs in City followed the standard process described above, the adoption of Restorative Practice throughout the local authority gave FGCs a central prominence in Children's Services work which is unusual within the UK. Social workers had to refer any family to the FGC service when an ICPC or prebirth case conference had been called, or there were otherwise plans to place a child in state care. As per the hybrid model used in City, the resultant FGC offer was non-mandatory for families. A family refusing an FGC would simply be subject to routine child protection processes. Equally, a family having an FGC would still also be subject to routine child protection processes if the family plan created at the FGC did not convince professionals that intervention should be de-escalated from CP level.

Levels of social care involvement in City
Firstly we compare how the changes in LAC, CP and CIN rates in City compare nationally over the period since Restorative Practice was introduced to the time the study data was completed. These statistics are point in time data taken on March 31st of each year (Tables 1 & 2) and show that the LAC rate in City, starting from a higher rate than average nationally, has been on a downward trajectory since 2012 when the use of Restorative Practice in City started to become widespread. By 2016, the LAC rate in City was still above the national average, however (76, compared to 60). Similarly, the CP rate has fallen consistently since 2013 to a level that has been below the national average since 2015. The CIN rate in City has been subject to greater variation and remained above the national average rate throughout the period from 2011 to 16. Nonetheless, it did decline from its levels in 2012, when the widespread introduction of Restorative Practice started to become embedded. At the same time, LAC rates fell by a fifth and CP rates closed to halved. By comparison, during the same period nationally, the LAC and CP rates rose consistently.
In order to contextualise how unusual City's trends in LAC rates have been during this period they were modelled over time and fitted for each local authority in England, with City's position in trends calculated relative to the mean trend and other local authorities outside of inner London (N = 136). Three model fit statistics are reported below: the Comparative Fit Index (CFI), and Tucker-Lewis index (TLI) and the Standardized Root Mean Square Residual (SRMS) which give an indication of the extent to which these more simplified trends are accurate reflections of the more nuanced data. We do not report the Root Mean Square Error of Approximation (RMSEA) statistic here due to its  sensitivity to small degrees of freedom (Kenny, Kaniskan, & McCoach, 2015). The most simple trend that could be fitted across the data was one that contained an intercept (the starting point of LAC rates at 31st March 2011), a slope (the average change over time for each additional year), and a quadratic term (which changes a linear prediction into a curve). A CFI of over 0.95 and a TLI of over 0.90 are usually considered good fitting models, whereas an SRMR of less than 0.08 is usually considered a close-fitting model (Kenny, 2014). By these criteria, the trends modelled offered a very good fit to the data. Initial levels and rates of change were allowed to vary for each local authority (Little, 2013). The model estimates for the trend lines for City were then extracted from the final model, and z-scores and ranks calculated for each component for comparison.
As Table 3 shows, local authorities in England (excluding inner London) had an average LAC rate of 61 children per 10,000 in 2011. In City, the LAC rate per 10,000 in 2011 was 94. This was approximately 1.4 standard deviations higher than the average, meaning only 8% of local authorities had LAC rates higher than City, ranking it as the local authority with the 12th highest LAC rates in 2010. While, on average, local authorities in England had LAC rates that increased at a slowing rate between 2011 and 2015, City was unusual in that it had a LAC rate that was decreasing at an accelerating pace: as time went on the LAC rate was falling at faster and faster speeds each year. Approximately 37.3% of local authorities had negative (falling), rather than positive (growing), slopes. Fifty seven per cent had negative quadratic change over time, meaning that year-on-year changes in LAC rates tended to move in a negative direction, such that if the LAC rate was increasing between 2011 and 2012, it would continue to increase, but increases in subsequent years tended to be smaller. Most local authorities tended to either have large increases in the start of the five year period observed that then 'flattened off', or had low or no increase at the start of the five years, followed by trends towards exponentially growing yearly increases towards the end of those years. City was different. The number of local authorities that had the same pattern as City was very smallonly 9 of the 136, or approximately 6.5%, of local authorities showed the same 'decreasing at an increasing rate' trend over time. In this sense, the local authority can be seen as within the normal distribution of change over time, but as an outlier of interest in terms of the shape that change over time has taken. It was one of very few local authorities that saw their LAC rate fall more and more as the years went on.

FGC service short-term indicators
The analysis of administrative data from the FGC service on six month LAC, CP and CIN indicators for families who had an FGC in City in the 2014 and 2015 years is now considered. These years were the only ones the for which the service had such data at the time of the study. The 2014 data covers 632 children, as 52 children could not be included in the analysis due to missing or restricted data. Data for 2015 had a number of children for whom data was either missing or restricted, or because it was less than 6 months since the FGC took place. The number of children on which the percentage calculations are based are indicated in brackets below each entry in Table 4.
The service indicator data show that relatively small proportions of children attending an initial FGC were in state care at that stage (11% in both 2014 and 2015), with the proportions slightly increasing to 16% after six months in 2014, and slightly decreasing to 10% 2015. Just under a fifth of children were pre-birth referrals in both 2014 and 2015. Pre-birth referrals are amongst the most likely to result in a child entering state care as the 'pre-birth' nature of the child protection referral reflects serious pre-existing concern about parenting capacity which is evident before a child's birth. This most typically arises when a parent has previously lost care of an older child or children. Research on recurrent care proceedings by Broadhurst et al. (2015) found a quarter of mothers were involved in repeat care proceedings within seven years, suggesting that having one child removed by the state is associated with a substantially increased likelihood of future state intervention to remove subsequent children. In City, the proportion of pre-birth children entering care did more than double by the six-month post-FGC stage in both 2014 (from 6% to 14%) and 2015 (from 5% to 12%). If we exclude pre-birth referrals there was a slightly smaller increase in the proportions of other children who were in state care at the six month post-FGC stage in 2014 (rising from 12% to 14%, rather than from 11% to 16% with pre-birth referrals included), and a slightly bigger decrease in 2015 (falling from 13% to 10%, rather than from 11% to 10% with prebirth referrals included). The proportions of pre-birth referral children in state care at the six-month post-FGC stage (14% in 2014, 12% in 2015) does indicate that the vast majority of children in this 'high risk'  City   2010  58  93  35  36  341  364  2011  58  94  39  64  339  407  2012  59  95  38  58  326  383  2013  60  87  38  63  332  417  2014  60  86  42  47  346  442  2015  60  78  43  38  337  401  2016  60  76  43  35  338  351 Sources: DfE, 2010a;2010b;2015a;2015b;. These are also the sources for Fig. 1-3 below. 1 All numbers and rates in the tables are taken from local authority returns to Government on numbers in those categories on March 31st of the given year. England CP Rate City CP Rate Fig. 2. CP Rate/ 10,000 Children. group were being maintained in their birth families in City in the early months of their lives. These proportions would seem high given the wider data on the likelihood of repeat removals occurring (Broadhurst et al., 2015). In 2014, the proportion of children in City who 'de-escalated' to having no social work involvement six months after the FGC was 17%, but close to doubled to 30% in 2015. The numbers of children classified as CIN in City showed the largest fall of all years from March 31st 2015 -March 31st, 2016, a decrease of 719 children (see Table 1). It would seem a reasonable claim that children who had an FGC were a notable part of this broader decline in CIN rates in City by 2016. Early Help services had been reconfigured in City in 2014 with a focus on de-escalating families from social work intervention and this may have been associated with the greater reduction in families subject to social work intervention in the 2015 year. If so, this would illustrate the importance not only of the FGC process itself, but those services available to families outside the FGC, in influencing whether social work involvement de-escalates.
In both 2014 and 2015, there was a decline in the proportion of children subject to CP six months after the FGC (11% fewer children subject to CP at the six month point in 2014, and 15% fewer at the six month point in 2015). However, the reductions in CP levels were smaller than might have been expected given the large declines in all children subject to CP in City between 2014 and 2016 (see Section 4.1 and Tables 1 & 2 above), and prior findings about the impact of FGCs on this indicator. Marsh and Crow's (1998) examination of FGCs in the UK reported that over half of children who were subject to CP were no longer six months after the FGC. It is notable that the decreases in the proportion of children without any social work involvement by the six month post-FGC point were greater than the decreases in the proportion of children subject to CP in both 2014 and 2015. One explanation for this difference, supported by some of the qualitative data from the ten case study families, was that where a family plan provided sufficient reassurance to professionals about a family's ability to address child welfare concerns, social workers moved to de-escalate and terminate their involvement swiftly. Where the family plan did not satisfy professional concerns, and children then became or remained subject to CP, it was difficult for parents to subsequently quickly convince professionals that children were not at risk of significant harm within the routine child protection case management processes to which they were then subject. That the family plan arrived at in an FGC did not always sufficiently abate professional concerns is likely to have partly reflected the greater seriousness of some families' difficulties compared to those families where children were de-escalated quickly following an FGC. However, it may have also partly reflected the tensions inherent in running a hybrid system whereby family plans constructed by families at an FGC had to run alongside, and possibly vie against, professional assessment of concerns and professionally driven processes.

Discussion
Following the introduction of Restorative Practice, City saw notable reductions in the numbers and rates of children in care and subject to CP from 2012 to 16. The reductions in the LAC rate were highly unusual compared to national trends. The six month FGC service indicator data show that the proportions of children in state care six months after an FGC were similar to the proportions at the time of the FGC, while the proportions subject to CP and with any social work involvement notably decreased by the six month point, particularly in the 2015 year. As the six month service indicator data do not have comparisons for those children who did not receive an FGC in City, it is not possible to estimate more exactly the extent to which receipt of an FGC was connected to the broader decreases in social work intervention rates in City from 2012 onwards. However, it was clear from the broader study (Mason et al., 2017) that the offer of an FGC as part of a Restorative Practice approach was an important marker of, and an important mechanism for operationalising, the local authority's commitment to supporting families' care of their children.
The six month service indicator data for children having an FGC do bear fairly close comparison with the six month service outcomes reported by Thoburn, Lewis, and Shemmings (1995) for 220 children on the CP Register in 1990 and 1991 in seven local authorities at the time the Children Act 1989 was being introduced. None of the seven authorities in the study of Thoburn et al. (1995) were using FGCs but they were all committed to the principles of working in partnership with families and de-escalating social work involvement where possible. It may be the case therefore that families receiving an FGC in City under the model of Restorative Practice had a similar experience of de-escalation from social work involvement as those studied by Thoburn et al. (1995) in an era, in the early 1990s, when LAC and CP rates were far lower than they have been since 2010. However, the differences in samples, methods of practice and the differences in time between the two studies mean this comparison must be treated with strong caution: it is raised here to illustrate the possibility that the use of Restorative Practice in City may bear some similarities to local authorities' attempts to work in 'partnership with parents' at the inception of the Children Act 1989.
It may be asked whether LAC and CP rates in City should be expected to fall further, given the LAC rate remained above the national average in 2016. The latest census data for 2018 showed that the LAC rates in City and nationally have indeed converged further over the last two years, but due to the national rate continuing to rise while the rate in City has remained constant (Department of Education (DfE), 2018).  The balances to be struck between working in partnership with families to maintain children in family care and ensuring that children remain protected have been a pre-eminent challenge in child and family social work practice internationally since the 1970s. In key informant interviews senior managers in City emphasised that the reduction in social work intervention rates had been achieved without leaving children in unsafe circumstances, a claim which seemed to be supported by positive inspection reports by the national inspectorate, Ofsted, on their services in the period since 2012. The operation of a hybrid system of FGCs did however mean that families were also still subject to the routine, investigatively driven, processes which characterise the English child protection system and of which most parents have negative experiences. In recognition of this tension, in 2016, City started to offer families an FGC in place of, rather than as well as, an ICPC. However, it was too early to evaluate the impact of this offer during the currently reported study. Further exploration of whether the embedding of FGCs as an alternative, rather than ancillary, decision-making process is more successful at reducing social work intervention levels would be of value.
There is also a question regarding which families are willing, and have the capability, to take advantage of the offer of an FGC, and which are not. This is currently poorly understood internationally. Future research increasing our understanding of why those families who do not take up an FGC refuse the offer at different stages of the FGC process would also be of value. We do know that care rates themselves are strongly associated with levels of family deprivation, the more deprived a family the more likely a child is to enter state care (Bywaters et al., 2018). It would be useful to know whether the likelihood of family take up of an FGC is itself linked to family deprivation levels or whether other factors, such as the size and strength of available family network support, or FGC co-ordinator practice, appear to be more influential.
Finally, it is worth acknowledging that the growing calls for child protection reforms have been premised on the assumption that reducing social work intervention levels is a desirable aim (e.g. Bilson & Hunter Munro, 2018;Gupta, Featherstone, Morris, & White, 2018;Trowler, With White, Webb, & Leigh, 2018). Indeed it is an assumption which has been tacitly adopted in this article and which derives from international concern about rising child protection referral rates and rising rates of state care for children. The assumption should not go unquestioned, however. Increased levels of social work intervention can arise from more positive underlying factors, such as the better availability of good quality care placements in a local area, or the use of short-term foster care as a form of family support, as envisaged when the Children Act 1989 was introduced in England. Decreased intervention levels can, by the same logic, arise from the absence of appropriate state resources devoted to meeting the needs of marginalised children and families. In the current socio-economic climate such an absence may be linked to an ideologically neoliberal agenda of greater welfare provision retrenchment and conditionality. In order to achieve enduring positive systemic changes to child welfare systems that will benefit families in the longterm a 'both/and' position is needed where arguments for reducing the involvement of coercive higher tariff state intervention sit alongside calls for state funding of other forms of welfare support, family support and youth work. Families experiencing significant difficulties need access to help which can support them to address those difficulties without that automatically resulting in them becoming subject to the surveillant gaze of child protection intervention.

Limitations
The findings related to national data in the paper should be interpreted with some caution particularly the comparison of CP and CIN rates: local authorities operate on differing interpretations of thresholds and there can be drastic changes between government administrations as local authorities react to changes, such as those made to funding formulas. The findings in respect of FGCs should also be approached with caution for three main reasons. Firstly, while the FGC service data for the 2015 year were available for a large majority of families, they were incomplete as six month indicators for families who had an FGC later in the 2015 year were still being collected by the time the study finished. Secondly, while both Government and FGC service data regarding social work intervention levels may be viewed as proxy indicators of wider child outcomes, they are short-term indicators and likely to provide limited representation of longer-term child, and indeed family, well-being and happiness. Finally, our analysis of administrative data related to FGC outcomes is unfortunately restricted to largely descriptive indicators due to data access restrictions. While we are able to provide descriptive statistics related to outcomes associated with FGCs, access to child-level data was not able to be negotiated due to its sensitive nature, and a number of variables of interest that we would have liked to control for were either not recorded or would have required data linkage to children's case files. As this could not be negotiated at this point in the study it was not possible to run more complex analyses controlling for confounding factors. The analysis here covers the entirety of cases within City, and therefore we can be fairly confident about the differences that appear. However, these do not represent a random selection of cases of FGCs outside of City, and therefore, in addition to the limitations above, the application of any statistical significance testing would be inappropriate for the purpose of attempting to generalise outcomes of FGCs to their use in all child protection cases. These differences should therefore be treated as illustrative of the case-study example of City and as possible avenues for exploration in studies with greater access to data and designs that are appropriate for generalising beyond a case study example.

Conclusion
Data regarding the reductions in the rates of children in state care and subject to Child Protection intervention in City paint a positive picture over a five year period from 2012, after Restorative Practice had been introduced. These reductions occurred in a national context where both rates were rising and were achieved at the same time as available indicators from the national inspectorate suggested the quality of Children's Services in City was good, supporting managers' contention that the reductions had been achieved without leaving children in unsatisfactory living arrangements. The exact contribution of FGCs to these declines is impossible to know with certainty from the available data. However, the short-term service indicators for children who had an FGC were potentially suggestive of partnership working with families reminiscent of an earlier era of lower state social work intervention rates. It can be concluded that the offer of an FGC to families, within a Restorative Practice approach, did indeed provide families with some opportunity to push back against state intervention. This in itself brings into focus the need to know more about which families are able to take advantage of the opportunity an FGC presents, and why, and which families are not, and why. The tensions in using a hybrid system of FGCs within the systemic constraints of a procedural model of child protection are also suggested through this study and raise the question of whether the use of FGCs as an alternative, rather than as an ancillary, to routine Child Protection processes may be more beneficial. Finally, the emergence of a growing international movement for child protection system reform raises the need for further debate about what policy and value considerations are driving, and should drive, the objective of reducing state social work intervention levels, and to what ends.

Funding
The study on which this article is based was funded by the Department of Education, UK, as part of the Children's Social Care Innovation Programme.

Declaration of Competing Interest
There is no declaration of interest for any of the authors which requires to be made.