Courts, care proceedings and outcomes uncertainty: The challenges of achieving and assessing “good outcomes” for children after child protection

The professed aim of any social welfare or legal intervention in family life is often to bring about “ better outcomes for the children. ” But there is considerable ambiguity about “ outcomes, ” and the term is far too often used in far too simplistic a way. This paper draws on empirical research into the outcomes of care proceedings for a randomly selected sample of 616 children in England and Wales, about half starting proceedings in 2009 – 2010 and the others in 2014 – 2015. The paper considers the challenges of achieving and assessing “ good outcomes ” for the children. Outcomes are complex and fluid for all children, whatever the court order. One has to assess the progress of the children in the light of their individual needs and in the context of “ normal ” child development, and in terms of the legal provisions and policy expectations. A core para-dox is that some of the most uncertain outcomes are for children who remain with or return to their parents; yet law and policy require that first consideration is given to this option. Greater transparency about the uncertainty of outcomes is a necessary step towards better understanding the risks and potential benefits of care proceedings.

but does not in itself guarantee emotional, social, or physical wellbeing. There are further questions about correlation and causality, the impact of wider social and economic factors, and the point at which outcomes should be assessed-perhaps, they are not truly known until adulthood? So what do we mean by "outcomes" and how can or should they be measured?
This paper considers some of the ambiguities and challenges of achieving and assessing "good outcomes" for children who have been subject to child protection court proceedings. The starting point is empirical data from a study of the process and outcomes of care proceedings for children in England and Wales. The full sample comprised 616 children, about half of whom were subject to proceedings starting in 2009-2010 and the others in 2014-2015-the first group before, and the second after, notable reforms to the care proceedings system (described below). Children in both cohorts were tracked for at least a year after their court case finished, whilst children in the first cohort were followed for up to 6 years. In the present paper, however, the focus is not on the statistical findings themselves but rather the profound issues that the data raise about the complexities of outcomes and well-being in children's social care.

| Outcomes and well-being in children's social care
Public scandals and media headlines often paint a rather bleak picture of the effectiveness of social work intervention to protect children from harm or to promote the well-being of children who are in care.
They tend to focus on the tragic cases of death or injury or, for children in care, the cases with numerous placement breakdowns, poor mental health, poor educational attainment, and poor adjustment to independence and adult life. Certainly, there are distressing cases, but research studies consistently show that there are also success stories, sometimes from starting points of great adversity and that there are often ambiguous and nuanced outcomes for individual children, with a mixture of positives and negatives. Useful research summaries are given by Forrester, Goodman, Cocker, Binnie, and Jensch (2009), Thoburn and Courtney (2011) and Boddy (2013).
Studies over many years have shown not only the challenges of achieving good outcomes for the children but also that local authorities do genuinely try to implement court-authorized care plans, whether those are for reunification with parents, kinship care, long-term care, or adoption. Studies regularly emphasize the importance of good assessment, planning, and support for parents and carers. In particular, they show the significant likelihood of returns to parental care breaking down and poor outcomes for the children in those that continue (e.g., Dickens, Schofield, Beckett, Philip, & Young, 2015;Biehal, Sinclair, & Wade, 2015;Farmer & Lutman, 2012;Harwin & Alrouh, 2017;Harwin, Owen, Locke, & Forrester, 2003;Hunt & Macleod, 1999).
As regards the well-being aspect of outcomes, English law gives specific requirements for children in care and involved with social care services, which have their roots in the "looking after children" materials of the 1990s. These were developed as part of an initiative to introduce the idea of outcomes to social work thinking (Ward, 1995;Parker, 1998); they identify seven key dimensions for considering a child's needs and progress, and the help they and their families may require. These are health, education, emotional and behavioural development, identity, family and social relationships, social presentation, and self-care skills. These seven aspects are now specified in the Care Planning, Placement and Case Review (England) Regulations 2010 (i.e., they are a legal requirement) as areas that must be covered in a child's care plan and considered whenever a child's case is reviewed (see DfE, 2015). Furthermore, the seven dimensions are also included in the "Assessment Framework," set out in statutory guidance for interagency working to safeguard and promote the welfare of children (DfE, 2018). This is a model for assessing whether a child is "in need" under s. 17 of the Children Act 1989, and/or suffering or likely to suffer "significant harm" (s. 47 of the Act).
The list of the seven aspects appears straightforward, but it is not hard to envisage that a child might be doing well in some areas and not in others (indeed, a mixture is possible even within each element-for example, reasonable physical health but poor mental health); or that a child's needs might fall primarily in one of the dimensions, but that affects their progress in others; or that progress might be variable over time, doing well at one point but not so well later, or poorly now but improving later.

| THE RESEARCH STUDY
The current study builds on earlier research by the same team (Masson, Dickens, Bader, & Young, 2013), which had investigated the use of "pre-proceedings" work by local authorities in 2009-2010, shortly after a new procedure for such work had been introduced. The process was intended to divert cases from court if possible or, if not, ensure that those that did go to court were better prepared, in order to reduce the duration of the court proceedings. The key finding from that study was that the new procedure had a negligible impact on court duration but did help to divert a significant proportion of cases (24%) through improvements in parental care, or alternative care arrangements (kinship care or "voluntary" care-accommodation under s. 20 of the Children Act 1989). The cases in this study were a random sample from six local authorities, and 290 children from the sample entered care proceedings during the research period. These children became Sample 1 in the present study.
From summer 2013, a new initiative to reduce the duration of care proceedings was rolled out across England and Wales, the revised Public Law Outline, introducing a deadline of 26 weeks for all but "exceptional cases." This time limit was subsequently incorporated into primary legislation in s. 32 of the Children Act 1989, coming into force in April 2014. It led to a dramatic reduction in the average duration of care cases, from 50 weeks in 2011 to 26 weeks in 2016, although it has risen again since then, up to 29 weeks in the first quarter of 2018 (MoJ, 2018).
But this change was accompanied by a number of high profile case judgments, which have also had a marked impact on local authority and court practice. There was a series of critical judgements about the "mis- It is widely seen that these lie behind a decrease in the proportion of care cases ending with adoption plans and an increase in those ending with kinship placements. The 26-week limit was not promoted as a way of changing the pattern of court orders, but it is impossible to say what the effects of the time limit alone would have been, given the impact of the judgments (Dickens & Masson, 2016;Beckett & Dickens, 2017;Masson, 2017). In addition, children's social care records for a purposively selected subsample of 118 children were examined. This comprised 10 children from each authority in each sample, except that, in one area, there were only eight cases for Sample 1. It represents about one third of the cases and one fifth of the children. The purposive sample was selected to ensure a mixture of ages and a range of orders and care plans but did not include children where the care plan was adoption and they had been successfully placed. The researchers also looked to select cases that had additional features that made them potentially informative. Examples are factors such as lengthy proceedings, cases where plans were strongly disputed or changed, or where the proceedings ended with complex contact plans for parents or siblings. The case file survey aimed to enhance the information available from the DfE data, by uncovering what happened in terms of children's well-being, the reasons behind any moves or changes of plan, and the services offered (or not) by the local authority and partner agencies. The researchers rated the children's well-being 1 year after the final order (T1) and, for the children in Sample 1, 5 years after it (T2: see below for more details of the rating process). This means we can compare the two samples in terms of court orders and shorter term progress (up to a year), and for Sample 1, we can compare the children's well-being over time, at 1 year and 5 years.
There were also interviews with 56 key staff in the local authorities about policy and practice changes and two focus groups with judges.
Permission to access the data came from the research governance systems of the participating authorities, Cafcass, the DfE, and the Judicial Office. Ethical oversight was from the researchers' universities.
Permission to access the DfE data, court records, and case files was dependent on the researchers making satisfactory arrangements for the security of the data (e.g., using a secure online data storage facility and secure laptops; case identifying details to be kept separately from the database) and undertakings regarding the anonymization of cases (e.g., for the statistical data, small numbers to be suppressed; for the case studies, pseudonyms to be used and nonmaterial details to be disguised).

| Outcomes (1): Court orders
The most striking difference between the two samples was the drop in the proportion of cases ending with adoption plans, which fell by half, from 30% of all final orders in Sample 1 to 15% in Sample 2. This was accompanied by a near-doubling in the proportion of cases ending with plans for children to live with relatives or other "connected persons" under special guardianship orders, up overall from 13% to 24%. The proportion of cases ending with the child returning to/remaining with one or both parents rose from a quarter to just under a third (25% to 32%), whereas the proportions ending in care orders stayed nearly the same, at about 30%. Within those figures, though, there were notable differences between different local authorities: for example, in one area, the proportion of adoption plans actually rose (although from a small base), and in another, the proportion of SGOs fell.
The children with adoption plans in Sample 2 were younger and were placed much more quickly in their prospective adoptive placements. Nearly all the kinship placements were still continuing at the follow-up points, but there were carers who were struggling to deal with practical matters (housing and finances), relationships with the child's parents and other relatives, and with the child's needs, as he/ she grew older. For the children in foster care, there were often long periods of stability, although it was harder to achieve this for older children. The placements that were least likely to endure were those with parents. Children subject to supervision orders only (i.e., not alongside a special guardianship order) were the most likely to have further care proceedings: New applications were made on 31% of those cases in Sample 1 (over 6 years) and 22% in Sample 2 (over 2 years). For fuller details, see Masson, Dickens, Garside, Bader, & Young, 2018a, 2018b, 2018c

| Outcomes (2): Well-being
In terms of well-being, the researchers assessed children in the purposive sample, using criteria devised by Farmer and Lutman (2012) in their seminal study of children returning home from care. These criteria enabled "researcher ratings" of the child's well-being at 1 and 5 years after the proceedings (T1 and T2). Well-being was assessed in a number of respects: the child's health, educational progress, any emotional or behavioural difficulties, peer relationships, relationships with current carers, relationship and contact with parent/s if the child was not living with them, family and social relationships, their social skills and social interaction, and finally, a rating for their overall well-being.
There were four main categories for this: "good," "satisfactory," "poor," and "very poor." There was also a category for not known/cannot tell.
The two field researchers scored the cases that they had each studied, based on their reading of the case file: The two lead investigators then rated the cases on the basis of written summaries provided by the field researchers. This gave each case three scores for overall well-being.
There was a total of 169 ratings (117 at T1: 58 + 60, minus one not known; 52 at T2: 58 minus six not known). The broad brush categories that we adopted from Farmer and Lutman (2012) worked well for this method: There was full agreement in 82, almost half. In nearly all the other cases, two of the scorers agreed and the other's score was only one grade different (e.g., two gave a rating of good and one satisfactory).
In only four instances were there three scores (i.e., good, satisfactory, and poor), and these were resolved by discussion. (There were few "very poor" ratings, so for the analysis we combined poor and very poor.) This was a purposive subsample, not a random one, and therefore one has to be cautious about inferring any wider conclusions about the well-being of the whole sample; furthermore, we only have T2 (5 years) ratings for the children in Sample 1. But focusing on the 58 cases from Sample 1, we found a fall between T1 and T2 in the number of those doing well, and an increase in those assessed as poor/ very poor. This applied to each of the three main groups in the purposive subsample-children on care orders with plans for long-term care (20), children living with kinship carers (12), and children living with one or both parents (19). (There was also a smaller group of seven children where the proceedings had ended with care and placement orders-ie, adoption plans-but from the information available they had not been adopted by the T2 checkpoint.) This sounds a discouraging finding, but it is important to look in more detail at the trajectories.
The least successful plans appear to be for children living with their parent(s). Of the 19 children in this group in the Sample 1 purposive subsample, four had been removed from parental care by T2, for adoption. (Two of these children's well-being had been rated as good at T1.) In all, there were eight children in this group whose well-being was rated as good at T1, but that had fallen to three by T2, for those whose placement continued that far. There was only one child whose placement continued who had a higher well-being rating at T2 than T1, but there were five rated as poor at T2, compared with one at T1.
For the children in the Sample 1 subsample in kinship care, 11 of the 12 cases were rated good at T1 but only four at T2. One placement had ended, and three were not known (all of these four cases had been rated as good at T1). The child whose placement had ended was in foster care at T2, and his well-being assessed as very poor. Of the other eight, the well-being ratings of three had gone down, whereas five had stayed level-four at good and one at satisfactory.
Of the 20 children in the Sample 1 subsample with plans for longterm care, 13 were rated good and two poor at T1; by T2, there were 12 good and six poor. Eight of the children had a fall in their rating, but five had gone up, and seven had stayed level, all at good. So although eight had experienced a fall, 12, more than half, had either an improvement or on-going good well-being.
Looking in more detail at the cases in the purposive sample helped to shed light on these trajectories and made it possible to identify a number of key challenges for achieving and assessing the children's longer term well-being.

| Challenges of achieving and assessing wellbeing
3.3.1 | So much is related to what has happened before, and to "normal" child development Nearly all the children in both samples came from extremely troubled backgrounds. It was unusual to find cases where there was only one specific, identifiable problem. There were more likely to be multiple problems, typically involving parental drug or alcohol misuse (including exposure in the womb), interpartner violence, and parental mental illhealth; other widespread features included parental physical ill-health, parental learning disabilities, chaotic and unhygienic households, sparsely furnished homes, poor diets, emotional abuse, and neglect.
The children themselves might have special needs or have experienced caring for younger children or their parents. Older children entering care were likely to have experienced many adversities, often over a considerable period. In these circumstances, challenging behaviour and/or poor mental health of the children is not surprising. High levels of on-going support for the children and their carers may be required.
For younger children, health and behavioural difficulties may not be known at the start but only become apparent later. This means they might have a good well-being score at T1, but by T2, things are more problematic. But this apparent deterioration should not necessarily be ascribed to "poor care"; it is more simply a result of symptoms emerging over time. In particular, the extent of special needs may not be known when children are very young and/or at start of proceedings. And all children are likely to go through some difficult periods during their adolescence, so some setbacks in (say) educational progress or emotional well-being during the teenage years could be regarded as "normal." The occurrence of the difficulties is less remarkable; it is the depth, extent, and persistence that are more indicative of serious concerns; and again, the difficulties may well have their roots in what the child has experienced before coming into care. The emergence of serious problems in adolescence for a significant proportion of adopted children has been noted in other research, even though only a small number of placements actually disrupt (Selwyn, Wijedasa, & Meakings, 2014).

| Balancing positives and difficulties, in different aspects of well-being
Progress in the different dimensions of well-being can be ambiguous.
A prime example concerns family contact. This can help meet the "family and social relationships" and "identity" aspects of a child's well-being, by giving them a fuller understanding of their background and reassurance that their parents or siblings are well and still hold them in mind, but it can be disturbing and disruptive. It is often not as simple as it being either positive or harmful, but rather both, simultaneously. There were cases where the comments and behaviour of the parents undermined the placement, both for children in foster care and children with kinship carers. As an example, in one of the cases from the file study, the maternal grandparents had to take out restraining orders against their daughter (a heavy alcohol user) in order to protect themselves from her, but their grandson enjoyed contact with his mother. By the end of our research period, however, the boy was now 14, still seeing his mother, but was now stealing from her, and the placement with his grandparents was in jeopardy.
Another example would be balancing stability in a placement against the different needs of different siblings. It is a legal requirement to try to keep siblings together, if that is consistent with the child's welfare and reasonably practicable (Children Act 1989, s. 22C), and this is a working assumption in social work practice. But on a practical level, there are some large families, with children of very different ages and different kin-relationships, and in such cases, it is likely to be hard to find a suitable placement where they can live together. The time spent looking for a suitable placement might affect the chances of it working well for all the children. In one example, the local authority spent over 2 years looking for an adoptive placement that would take two sisters, both aged over 4. They did eventually find one, and placed the girls together, but after 18 months, the prospective adopters asked that the older girl be moved because she was not settling into the family, was self-harming, and aggressive towards her sister. Two years later, both girls were reported to be doing well in their respective placements, and there was positive direct contact between them. Here, both children appeared to benefit from separate placements, but more challenging still are cases where different children have incompatible interests, thus raising a fundamental question: consistent with which child's welfare? Plans for children to stay together were somewhat prone to changes, then raising practical questions for social workers as to what on-going contact might be suitable (and again, what is suitable for one child may not be for the other).

| Well-being is variable; many different factors can change the picture; causality is not always clear
Reading the case files showed how much can change over time for the children. Assessing well-being at any point is only a snapshot of a particular moment. We assessed well-being after 1 year and again after five, but the files showed that things can change between and afterwards; and also, that the well-being scores might be the same, but the circumstances very different. As an example, one case involved a 2-year-old girl who was living at home with her mother at T1, under a supervision order. The researchers rated this "good" at T1. But the placement broke down 17 months after T1, and the girl came back into care. There were new care proceedings, and a foster care breakdown because the foster carers were not able to manage her behaviour. So had we assessed her well-being after 3 years, it is likely to have been "poor." By T2, however, the 5-year point, the girl had been adopted and was doing well in her new family. But it had been a difficult time, and the circumstances were very different now (and may change again: the consequences of this difficult time may come to be felt further in the future). In another example, a 9-year-old boy had been the subject of care proceedings because of neglect and risk of harm from his mother's poor mental health, alcohol misuse, and interpartner violence. The proceedings ended in a supervision order, and he went home. At T1, he was doing well. By T2 though, aged 14, he was not attending school, was held to have very low selfesteem, and was acting violently towards his mother. He was still at home, on the waiting list for support from specialist "child and adolescent mental health services" (CAMHS). His mother was experiencing many health problems. So at T2, his well-being was assessed as poor.
But the field researchers were able to read ahead in the file, beyond T2, up to the date that they were studying the file. A year after T2, things had improved considerably. He was now attending a special therapeutic school, doing well, and his mother was much better too.
Although good care provides the foundation for children's progress, many other factors can intervene. "Chance" can seem to play a large part, for good or ill. There were a number of examples of successful placements being disrupted because of carers unexpectedly becoming ill, or even dying.
But "good luck" could play a part too. One example concerns a troubled young man on a care order, who was in residential care at T1. Later, he went into foster care. He had two foster placements, one ending when he threatened the carer with a knife. But despite that setback, he was actually making progress and later got a job and restarted college. A crucial thing for him was that he got a girlfriend whose family was very supportive; he ended up going to live with them when he was 18 and was still in that relationship 3 years later.
Chance? Maybe, but there must have been something there that enabled him to respond positively to that family's support-and he had been helped to preserve that, or to rediscover it. One of the tasks of a parent or carer is to create the conditions for serendipity to happen (Gilligan, 1999).

| Sometimes limited information on which to judge
This was more likely for the children not in care, perhaps with parents or kinship carers, where cases had been closed, either because of a decision that the case no longer needed to be open, or because the family had moved away. Our study showed that a number of these families were subject to care proceedings again in their new authority.
It has to be said that information not always clear on the files of cases that were still open, and it was sometimes hard to track the reasons why decisions were made, or plans changed. That is not to say they were necessarily unjustified decisions; rather, the evidence and analysis behind them was not clear from the written records.
Young people and families may be reluctant to engage with workers and services, or ambivalent. "Tailing off" of engagement was not unusual, particularly for families where the children were living with their parent(s), but seemed more likely to happen after the first year. This is a challenge for the use of supervision orders, which last for up to a year in the first instance. They may be extended for up to 3 years, but if all seems well at the end of the first year, it is unlikely they will be; but it is after that, that difficulties are likely to emerge.
Young people might show their own feelings by "voting with their feet." In one case, a 12-year-old girl and her older brother absconded from foster care after a contact visit with their mother, 3 months after care orders were made, and returned to live with her. The local authority decided not to try to remove them, and although there were difficulties, the children were still there at T1, and their mother was now accepting help. In another example, there was a care order on a teenage girl, but she often went missing from her foster home, going back to relatives. Eventually, she stopped returning to the foster home at all, and it was not known where she was staying, but she was still attending school and doing well there.
Children might also play an active role in shaping what happens to them by saying they want to stay in a placement. But this is not necessarily straightforward. An example is a case where a boy strongly expressed his view that he wanted to be adopted by his foster carers.
The local authority social worker and independent reviewing officer both had doubts about this, but the boy and the carers pressed for it. Despite the reservations, he appeared to be doing well in the placement and the researchers assessed his well-being as good at T1. Eventually, the authority agreed to the plan, but shortly afterwards, the placement ended when the boy alleged mistreatment by the carers.
At T2, he was in another foster placement, and not doing well; but post-T2, he moved to a residential unit, where his behaviour was still causing concern, but he was attending school and doing well there.

| Resource limitations in children's services and other agencies
High levels of demand, financial restrictions and staff shortages in children's services and their partner agencies, were making it increasingly hard for agencies to offer support to children and families. Support at all levels was affected: early, preventive support for families, financial support for kinship carers, and therapeutic support for troubled children and their carers. Partner agencies were also being hit by cuts, and it was notably hard to secure timely input from CAMHS. Criteria and thresholds for CAMHS varied from area to area and could be very high-in one notable example, CAMHS refused to work with a child because they considered the case too complex. Three of the local authorities in the study had set up their own specialist services to deliver therapeutic help to children in care and kinship care.
Preparation for independence and transitions to adult social care services could be problematic, with disagreements about the nature of the young person's condition and which service was responsible (e.g., is the primary need for support for learning disabilities or mental health?). There were examples of good outcomes being jeopardized because of breakdowns in the arrangements for transferring to adult services. One example involved a young man who had been in the same foster care placement throughout his time in care, but as he approached 18, concerns grew about his ability to live independently.
It was hoped he could remain in the same placement, but it ended a year before T2, because of lack of funding from adult services. He moved back to live with his mother. A year after T2, it was recorded that he was not engaging with services and support from adult services had ceased.
Despite the challenges, there were examples of sustained, effective work from professionals (social workers, teachers, support workers, and mental health specialists) and carers. There were examples of parents, foster carers, and kinship carers being helped to meet the child's needs more effectively and of direct work with children to help them (e.g., life story work, helping children cope with bereavement, and helping children to attend and do well at school/college).
But "effective work" does not necessarily mean supporting the carers and child to stay in a particular placement: Sometimes, moving the child was necessary to secure improvements in their well-being.
Sometimes, there were difficult decisions to be made about how long to support a placement and the uncertainties of moving the child, and weighing concerns about some aspects of the child's well-being against other more positive aspects.

| DISCUSSION
Following children's progress after care proceedings highlights the commonalities and differences between different sorts of cases. There are children who are returned to parental care (or have never been removed from it), children with extended family, and those with plans for long-term care or for adoption. Almost all the children are likely to face considerable challenges and deserve good care and effective, sustained support. The child's well-being is the goal for all of them, but how this is understood and how it is to be achieved varies greatly (e.g., questions about contact with parents or siblings, as discussed above). In particular, the file study showed the nature of difficulties for children living with their parents, the risks of breakdown over time, and the challenges for kinship carers. These are not new findings but already well-known, echoing those of previous studies. But law and the "social contract" require courts and agencies to place (or leave) children with parents or kin if possible. If the primary goal is to keep children with their birth families, this may require some trade-offs in other aspects of their well-being; and these then become ethical and political questions, rather than just "technical" ones about child development or researcher ratings. This ethical and political dimension was sharply captured over 30 years ago, when Dingwall, Eekelaar, and Murray (1983: 244)  on the facts of the case but need a "frame of reference" help them interpret those facts, and research into the outcomes of care proceedings can provide one such frame (Masson, 2015). The present study shows how complex even this is, given the range and subtlety of outcomes.

| CONCLUSION
The questions raised in the discussion clearly have profound social, ethical and political dimensions, but they also have practical consequences. Four stand out. First, that social work (and other) assessments in care proceedings have to be comprehensive and accurate, not just to describe and explain what has happened in the past (although that is essential for showing patterns of behaviour) but to look into the future as to what this is likely to mean for the child's well-being, and what help he/she may therefore require. Second, for court decision-making, that it has to be realistic about the impact of past harms, both on the children and the parents, especially that parents are likely to find it hard to sustain the changes they may have achieved under the spotlight of care proceedings. The third message is that on-going support is likely to be necessary for all the children who have been through care proceedings and their carers (parents or others), and adequate resources need to be made available to local authorities and partner agencies, through government funding (i.e., taxes). Local authorities need to plan carefully for this on-going role and should be cautious about closing cases too quickly-although this has to be set against their workload pressures, the ambivalence of some of the families, and the fact that services cannot be imposed if there is no court order. Finally, the paper has highlighted the importance of realistic expectations about the limits of state intervention and the limits of predictability, that some cases may work out better than expected whereas others may fare worse. The judge's comment above is easily said in a courtroom in support of a decision he/she is making, but it then needs judges and others (politicians, professional regulators) to hold to it and resist the tendency to blame someone when things do not go well. This is not an excuse to deny responsibility for one's part in the decisions and any subsequent actions, but to recognize the intrinsic complexity and uncertainty of outcomes. Hard as it may be to acknowledge, these are not just about the well-being of the individual child but about the nature of legal decision-making and the values of society.