Four-year longitudinal impact evaluation of the action for children UK neglect project : outcomes for the children, families, action for children, and the UK

Neglect has a devastating impact on children and is the most pervasive form of child maltreatment in the United Kingdom. The study purpose was to establish outcomes for neglected children following structured assessment and intervention to ascertain what worked and why it worked. This prospective cohort study included 85 cases of neglected children under 8 years of age from 7 centers across the United Kingdom. Data were collected between 2008 and 2012 through serial quantitative recording of the level of concern about neglect. Serial review of qualitative caseﬁle data was undertaken for detail of assessment, interventions, and evidence of outcomes for the child. Data analysis was undertaken by paired t -test, Chi Square, descriptive statics for categorical data, and, for narrative data, identiﬁcation of recurring factors and patterns, with correlation of presenting factors, interventions, and outcomes. Paired t -test demonstrated signiﬁcant decrease in overall Action for Children Assessment Tool scores between assessment ( M = 43.77, SD = 11.09) and closing the case ( M = 35.47, SD = 9.6, t (84) = 6.77, p < 0.01). Improvement in the level of concern about neglect was shown in 79% of cases, with only 21% showing no improvement. In 59% of cases, concern about neglect was removed completely. Use of the assessment tool fostered engagement by par- ents. The relationship between lack of parental engagement and children being taken into care was statistically signiﬁcant, with a large effect size ( (cid:2) 2 10.66, df 1, p = 0.0001, OR = 17.24). When parents refused or were unable to respond positively to the intervention, children beneﬁted from an expedited move into care. © 2013 Elsevier Ltd. All rights reserved.


Introduction
Public inquiries repeatedly show that families of some neglected children fail to receive adequate services, sometimes with tragic consequences. Because neglect is the most common category for child protection registration in the United Kingdom (Department for Education [DfE], 2012;Department of Health, Social Services and Public Safety, 2012;Welsh Assembly Government, 2012), timely and effective intervention is vital. Successful intervention to improve parenting ability, establish appropriate intra-family relationships, and secure positive outcomes for children "is likely to be costly, requiring intensive, long-term, multi-faceted work by a highly skilled workforce" (Moran, 2009, p. 25). However, if these factors can be addressed, then positive outcomes are possible. Furthermore, recent evidence suggests that when neglect is entrenched and assessment indicates that the family situation is unlikely to change, then intervening to take children into care earlier rather than later improves outcomes in later (young adult) life (Hannon, Wood, & Bazalgette, 2010). Skilled, early intervention is, therefore, a key factor both in cases where neglect can be addressed successfully and in cases where there is no indication of likely improvement.
Research evidence guiding policy and practice is somewhat limited, tending to be based on studies of short duration that provide only snapshots of activity and outcomes. It is also mostly retrospective, focused on proxy measures of outcomes, and derived from small-scale studies (Daniel, Taylor, & Scott, 2009). Because the nature of neglect often demands longer-term intervention for sustained improvement to be achieved, then a longitudinal approach to evaluation is also indicated. This study was commissioned by the organization "Action for Children," which provides services throughout the United Kingdom to improve outcomes for neglected children. The aim of the study was to develop an understanding of what worked and the reasons for success in order to guide future strategy and practice.

Defining and understanding neglect
Neglect is complex and challenging to define (Moran, 2009), and variations in definition can create difficulty in the interpretation of research findings (Stein, Rees, Hicks, & Gorin, 2009). The working definition provided by the British Government was well-known at the commencement of the study and had been retained and reaffirmed as central guidance for several years.
The persistent failure to meet a child's basic physical and/or psychological needs, likely to result in the serious impairment of the child's health or development. It may involve a parent or carer failing to provide adequate food, shelter and clothing, failing to protect a child from physical harm or danger, or the failure to ensure access to appropriate medical care or treatment. It may also include neglect of, or unresponsiveness to, a child's basic emotional needs (Department of Health [DH], 1999, p. 6).
This definition was adopted and retained throughout the study. A revised definition in the following year (HM Government, 2010, p. 39) brought attention squarely onto the impact on the child and reinforced the issue of parental fault or intention not being necessary for the child to suffer neglect. However, these issues were already acknowledged in the project.
Growing understanding of the factors that contribute to neglect has led to the development of categories such as medical neglect, nutritional neglect, emotional neglect, educational neglect, physical neglect, and lack of supervision and guidance (Horwath, 2007) that help to pinpoint aspects of care that may be the focus of neglect. Indeed all of these featured in the referral, assessment, and intervention reported in the cases in the evaluation. Crittenden's (1999) consideration of the causes, manifestation, and response to neglect also offers three common presentations which include disorganized neglect, emotional neglect, and depressed neglect. The categorization of cases by cause or circumstances in this manner can be oversimplified and misinterpreted, but helping practitioners to include such analysis in their appreciation of the antecedents to neglect might promote better selection and targeting of interventions. Health visitors (public health nurses) have been found in this manner to recognize both the parental characteristics associated with neglect and the signs in children of developmental problems (Daniel et al., 2009).

Effectiveness of responses to neglect
Usually, families which come to the notice of safeguarding systems have shown signs of neglect over a number of years, and there is little evidence to suppose that short-term intensive approaches work in the field of neglect without long-term follow-up. Longer-term interventions seem to be more productive but are challenging to child care systems because neglect is characterized by repeated need for intervention, with families requiring long-term support (Davies & Ward, 2011;Moran, 2009). The level of need creates significant demand on practitioner time, interagency collaborative effort, and emotional energy.
The complex series of relative judgments associated with neglect cases often leads to a lack of agreement on the threshold for intervention, particularly at the threshold of removal into care (Platt, 2006;Stevenson, 2007). Neglect may also overlap with other forms of maltreatment, and although recognition of physical or sexual violence may lead to a decision to escalate intervention, it is sometimes the day-to-day neglectful interactions that lead to the most harm to the child (Platt, 2006). Table 1 The continuum of need and response in UK Child Welfare Policy.

Adoption
The best interests of the child require permanent removal and adoption. 6 Care proceedings The child's health and development are at such risk that the case is put before the court with a view to parental responsibility being shared with the Local Authority and the child coming into 'care'. 5 Child protection plan A formal, multi-disciplinary plan is made to safeguard the child's welfare. Parents are involved, the child remains in their care, but there are clear objectives and deadlines for review. 4 Section 47 Inquiry (of the Children Act 1989) If a child is at risk of, or already suffering, significant harm, then the Local Authority has a duty to undertake a formal, multi-agency inquiry into the child's needs and best interest, and to act to protect the child. 3 Child in need A child and their family have complex problems that need Local Authority intervention in order to meet acceptable standards of health and development. Multi-agency involvement will be required.

Thresholds of need
A threshold model of need has been incorporated into most child care systems in the United Kingdom with some variation across its constituent countries. The model is based on the notion that children reside on a continuum of need, and that health and social care services should be provided on a matching continuum. Seven categories are conceived, from universal services (which every child should access) to adoption of children through the court and formal legal systems (see Table 1). From level 4 upwards, the case is pursued through statutory processes. Much of this structure and process was originally based on the Children Act 1989 and has been further developed through the Every Child Matters series of Government policy statements from 2003 (now decommissioned) which culminated in the Children Act 2004. Although this categorizing had been adopted some time before, many relate it readily to Hardiker levels (see Hardiker, Exton, & Barker, 1991, for further details).
As an agreed level of concern that is sufficient to trigger a service response, the threshold for intervention for children who are being abused or neglected has been the subject of debate for many years in the United Kingdom and elsewhere (Platt & Turney, 2013). It is now widely recognized that a range of factors affect thresholds including policy and organizational circumstances, collaboration among professionals, and the nature of the concerns.

Why neglect is neglected
Assessment. Assessing a family's private domain is always difficult. The British safeguarding system was originally established to measure acts of commission, particularly around physical or sexual violence. However, assessing neglect clearly involves measuring a complex series of acts of parental omission and their subsequent impact on the child's development and wellbeing. The true impact of neglect may not become apparent for some time, and the child can be many months older before the system gets to grips with the family situation (Horwath, 2007;Platt, 2006).
Analysis. Variations in definition and the responsibility for making judgments about highly complex and chaotic family situations, sometimes in the context of changeable thresholds, can make the analysis and decision-making process especially complicated (Macrory & Murphy, 2011). Guidance on judgments about acceptable levels of parenting in families is offered in the Children Act 1989, which refers to significant harm and reasonable parenting, and in a more recent interagency policy report (HM Government, 2010, p. 39), expands on persistent [parental] failure to meet a child's basic needs and serious impairment of health or development. A series of extra tools, including the Home Inventory (Cox & Walker, 2002) and the Graded Care Profile (Pollnay & Srivastava, 2001) have also been developed to assist in this process. However, such tools depend on the practitioner's ability to analyze what they see of the family's private domain and to translate that into complex decisions about what is and is not child neglect.
Action. The collective action required to help families to change is also potentially problematic. The British safeguarding system is set up to deal with the collaborative "short sprint" whereas neglect involves an interagency "marathon" where practitioners and systems have to act collectively and respond to families over a number of years. This poses ongoing challenges to child care systems (Hallett & Birchall, 1992). Reder and Duncan (2001) noted that intergenerational distress, or the failure of attachment between parent and child, can be sustained and replicated from one generation to the next. With parents who are persistently neglectful, notably those with substance misuse or domestic violence problems, children and practitioners may be prone to feelings of disappointment and anger at the lack of progress and repeated relapses (Hart & Powell, 2006). These feelings can be seen as a manifestation of the emotional labor of working with neglectful parents and the sustained effort and commitment required to pursue matters to a satisfactory conclusion draws heavily upon professional emotional resilience (Social Work Reform Board, 2010).
Adult-orientation. Substance misuse, parental mental ill-health, domestic violence, and learning disability can often be significant in cases of neglect, as they can mean the parent is less available and responsive to the child. Child care staff may be unfamiliar with measuring these issues or calculating their impact on the child (Cleaver, Nicholson, Tarr, & Cleaver, 2007;McCarthy & Galvani, 2010), so the adult-oriented issue can become the prime focus of attention. This challenge makes the interagency collaboration to address the child's outcomes more complex.

Summary of factors in neglect interventions
The challenges to achieving the requirements of intensive, long-term, multifaceted intervention, particularly in a persistently difficult financial context, are enormous. Although previous reviews of evidence to inform practice for Action for Children identify promising aspects of intervention (Moran, 2009;Tunstill, Blewet, & Meadows, 2008) including home visiting, parent training, school-based social worker support and intervention, social network support, and therapeutic approaches with parents and children, the caliber of evidence reviewed is variable.
A review of child neglect by Burgess, Daniel, and Scott (2012) demonstrates that the current structure of child protection systems in the United Kingdom can militate against effective action on neglect. They contend that increasing financial pressures are impacting severely on children's wellbeing, which prompts them to call for a commitment by the United Kingdom and devolved administrations to a longer-term approach to intervention for neglect. Improving clarity among professionals and the public over what constitutes neglect would also help to ensure earlier and more effective intervention.

Study design
Already cognizant of the limited results achieved by short-term, snapshot reviews, Action for Children commissioned a longitudinal study over four years from November 2008 to July 2012.

Selection of sites and cases
Data were collected from a group of seven services, with sites in Scotland, Wales, England, and Northern Ireland. These services varied in the range of cases addressed, though all received cases with concern at levels 4 and 5. All seven services received direct referrals from the Local Authority (local government with responsibility for children's services), and two also received self-referrals, though these were not included in the study. Two sites enjoyed particularly positive relationships with health professionals and received referrals from health visitors. Fixed-term contracts for interventions were usual, often of around three months, though sometimes extended when progress was shown. In other cases, families were returned to previous services for continuation of support and intervention, though often with remarkable progress already made. Four of the sites included widespread areas of poverty, often linked to unemployment. The severity and number of problems identified was correspondingly higher in these centers.
A continuum of focus was observed, from an entirely outreach-based service to regular group work in-house, but most services engaged with families through both approaches. The majority of workers were not professionally qualified, but they were supported through supervision and case review by local managers. Workshops were held before the sites became involved in the study and at 6-month intervals thereafter to deal with practical issues of data collection and analysis. Monthly meetings with all site managers further added to uniformity in approach and use of the instrument.
Eighty-five cases were included, with a further 10 lost to missing data. The inclusion criteria were that the case related to a child under the age of eight years, neglect had been identified explicitly by a referring agency at the point of referral or was identified as an emerging concern on assessment, and the Action for Children Assessment Tool was used and summary scores recorded. Excluded families received the normal package of care. If one or more children met the inclusion criteria, then the family was included, but data were collected about the eligible children only.

Data collection
The evaluation was based on quantitative recording of the level of concern about neglect in specified areas at least on referral and on closure; electronic recording of key characteristics of the child, the parents, and the environment; and review of qualitative textual data from case files detailing issues on referral, specific interventions, and evidence of outcome for the child. Serial review of the files and scores allowed for ongoing recording of progress, or lack of it, in each case.
Action for Children Assessment Tool. The assessment tool used in this study was based on revision of an earlier tool, itself derived from previous tools. The Action for Children Assessment Tool is divided into three general areas of concern: physical care, safety and supervision, and emotional care, each populated by a number of specific elements (14 in total; see Table 2). The items are scored from 1 to 5, with higher scores indicating increasing levels of concern. A threshold was set at the  Safety features Health boundary between 3 and 4 where scores below this line (1-3) represented at least adequate care. Scores of 4 or 5 indicated inadequate care and serious concern. This threshold was an essential element, as it provided practitioners with a means to gauge which aspects of care were unacceptable (rather than simply undesirable) and which to prioritize for intervention. Each item bore descriptors at each level to assist the practitioner further. Serial scoring of items allowed for clear indicators of overall progress and lowering of concern. Recording the number of scores of 4 or 5 offered the research team an additional index of the degree of concern and the direction of movement.
A number of other assessment instruments were used routinely, and the information from these was also incorporated into the action plan. Referrals were made to other agencies as required (substance misuse, housing, crèche and nursery, and health services), and the assessment and interventions from these were also factored into the plan of care.
eAspire. Action for Children also maintains an internal electronic system (eAspire) which tracks the child's progress through the service from referral to closure of the case. For the purposes of the study, custom fields were added to this in order to collect data pertaining to the child's health characteristics; the child's education and emotional wellbeing; parent/carer characteristics; environmental factors, presenting needs, and interventions. The outputs from this helped to show which factors were most frequently reported and which commonly presented in combination.

Data analysis
The overall effect of the intervention (reduction in total Action for Children Assessment Tool scores) was analyzed using a paired t-test. Although the scores themselves were not normally distributed, Field (2013) suggests that the assumption for the use of this test is that it is the distribution of the changes between the two sets of scores that is critical. In this case, the difference was approaching normality.
For categorical data, analysis was undertaken by nonparametric statistics (Chi Square) when possible; otherwise, descriptive statistics were applied. The qualitative textual data was derived from commentary written by professionals in various documents that constituted the case files which was analyzed using framework analysis (Smith & Firth, 2011). This process involved sifting, charting, and sorting data in accordance with key issues and themes using a five-step process of familiarization, identifying a thematic framework, indexing, charting, and mapping and interpretation. In framework analysis, once the researcher is familiar with the data, the emerging issues, concepts, and themes expressed by the participants can form the basis of a thematic framework which facilitates the filtering and classification of the data (Srivastava & Thomson, 2009). The framework for this data included the home, the parents, and the child, and the data within this frame was used to help to explain and illustrate the numerical findings.

Ethical issues
Action for Children had written permission from service users to use personal data for the purpose of research-driven service improvement. Information sheets and consent forms were, nevertheless, tailored to each site as required by the local manager. All data was anonymized and a case number applied to each child's data. No identifying information was included in reports or dissemination activities. Data was stored on a password-protected computer in the university, backed up on a secure server drive. Hard copies of materials were stored in a locked filing cabinet in the chief investigator's office. Formal approval was secured from the University of Salford Research Governance and Ethics Committee.

Population-level electronic data
Child, parent/carer, and environmental indicators. The condition of the child's home is fundamental to the assessment of wellbeing and parenting capacity. Ferguson (2009, p. 427) suggested that in past practice, aspects such as the child's sleeping conditions were considered to point to the "truth" of the family's inner life and the child's welfare, but little is known about whether there is routine inspection of the home by social workers today. In this study, the Action for Children instrument gave workers an opportunity to consider home conditions in detail, and a wide range of factors which had stimulated referral was identified. Chaotic lifestyles with no routines (n = 44, 52%) and poor home conditions (n = 35, 41%) were predominant problems and often occurred together. Poor hygiene (n = 29, 34%) was also commonly associated with these, and finding was also seen in analysis of the qualitative textual data. However, other issues were also widespread, with a significant degree of domestic violence (n = 29, 34%) known to exert a particularly negative impact on children's wellbeing (UNICEF, 2006). Failure to attend for health appointments (n = 15, 18%) and poor hygiene (n = 17, 20%) were the most commonly reported factors in children's health characteristics. However, relatively little focus was placed on factors relating to the child, which was a limitation that was mirrored in all aspects of the data. The higher incidence of emotional problems (n = 24, 28%) and behavioral problems (n = 20, 24%) noted in the child reflects the focus and level of need that was present in many cases. The problems identified in children could have been either a contributing factor or a result of the chaotic family situation.
The much greater prevalence of factors noted in parents meant that the main effort made by workers was focused on parental behavior. However, the degree to which these interventions address mental health, substance misuse, or learning disability is not clear (see Table 3). The 23 (27%) cases in which parental substance misuse was recorded invariably involved referral to local drug and alcohol services, while project workers focused on maintaining the child's safety. Little was recorded to indicate that environmental factors were a particular influence in the cases in this study. Child protection issues as either a child protection plan or court proceedings were common (n = 45, 53%).
Interventions. There was a predominance of home visiting and associated focus on routines and boundaries which the case files review revealed to be drastically underreported in the electronic data. The relationship between the application or not of home visiting (or outreach) and whether or not the child was taken into care was not statistically significant ( 2 0.783, df 1, p = 0.317, OR = 2.02). A larger sample size may be required to detect significant differences, though more accurate reporting of such data might have a more important effect on the results. Some items were a required part of the commissioned service, for example comprehensive assessment and parenting program (see Table 4). A specific parenting program (usually the Positive Parenting Program [PPP]) was often requested as part of the commission (n = 29, 34%), possibly because of perceived rigor associated with this package. However, a recent independent evaluation of the PPP found no significant difference in parenting stress, positive interaction, family functioning, or child problem behaviors between this program and other group-based programs (McConnell, Breitkreuz, & Savage, 2011). A slightly smaller frequency was reported for use of the Webster Stratton (Incredible Years) program (n = 22, 26%). The effectiveness of this program has been supported in Sure Start centers (Hutchings et al., 2007).
A national evaluation of such programs found that both programs could be effective, provided that there was fidelity to the program design, that trainers were adequately prepared and supported, and that there was a practical fit with local circumstances and organizational approaches (Lindsay et al., 2011). These provisos were met in the case of services included in this study. Importantly, with most evaluations, the measure of outcome for children is limited to behavior modification  as indicated by the Strengths and Difficulties Questionnaire (www.sdqinfo.com). This limitation highlights the need for inclusion of additional indicators of child outcomes if a more reliable evaluation of the success of interventions for neglect is to be achieved. In any case, both service users and practitioners identified home visits as the most important aspect of the service (in which the learning from parenting programs was applied to practice).

Action for Children Assessment Tool data
Removal of concern about neglect. Overall, the intervention was successful. A paired t-test demonstrated a significant decrease in overall Action for Children Assessment Tool scores between time of original assessment (M = 43.77, SD = 11.09) and time of closing of the case (M = 35.47, SD = 9.6, t(84) = 6.77, p ≤ 0.01). The eta squared statistic (0.35) indicates a large effect size. The reduction in all Action for Children Assessment Tool scores to below 4 was taken to indicate complete removal of serious concern about neglect (though concern about parenting generally might remain). In 50 (59%) cases the intervention was so successful that the cases were transitioned to mainstream services with no remaining concern about neglect on closure. Most of these cases transitioned into Child in Need or Targeted Services, though some were resolved to the extent that the families were handed back to universal services for ongoing support (see Fig. 1).
Typically, parents would have learned to maintain a clean, tidy, and hygienic house, and they would have established a more positive, stimulating relationship with the child by the time of the final review. Boundaries would have been set in place and maintained by the parents. Other common indicators were better school attendance, more effective and appropriate communications between parents and children, and greater awareness of hazards to young children. The achievement of even a small change could take a great deal of intensive intervention, and frequently steps would have to be retraced and learning fostered again before moving back to the timetable of progress. However, the persistence of the worker in these cases prevail, and parents commented on the positive effect of this persistence and stability on their own resolve to provide better parenting for their children.
Prevention of neglect. In a small number of cases (n = 8, 9%), scores never reached the threshold of neglect. These were mostly cases of young parents or parents with learning disabilities whose inability to provide adequate care was already of concern antenatally, and the intervention was aimed at preventing foreseeable, inevitable neglect while developing the essential skills in the parent. Each case focused on specific deficiencies in parenting. The remaining cases related to holding actions Change in total areas of serious concern (scores of 4 or 5) in the three major areas of the Action for Children Assessment Tool.
in anticipation of a foreseeable improvement to the family circumstances. For example, in one instance this related to the return of the children's father following his release from prison. The mother's care of the children was deficient in both knowledge and skill such as to be neglectful, but on the father's return, he was seen to be cooking for the children and enforcing reasonable morning and bedtime routines. In all cases, otherwise inevitable neglect was prevented, and in six of these eight cases there was improvement in caring ability rather than merely maintenance just below the threshold of neglect.
Persistence of concern about neglect. In 27 cases (32%), there was remaining concern about neglect. These concerns ranged from a single, lingering problem which would require longer intervention than was allowed in the commission to a complete absence of change or even worsening of the situation. Four cases were returned to the commissioner for continuation of the care plan. Twenty-three cases resulted in the child being taken into care on closure of the case. These cases tended to relate to families in which neglect was already deeply embedded upon referral. The cases were severe and complex. Commonly, little or no change would be seen in the degree of concern about neglect, sometimes with concern increasing as additional issues came to light. The most common reason for closing a case was that the parent persistently failed to engage with intervention or was simply unable to make significant improvement in their care of the child. The relationship between lack of parental engagement and children being taken into care was statistically significant, with a large effect size ( 2 10.66, df1, p = 0.0001, OR = 17.24). In these cases, the strength of evidence that the child was suffering neglect and that parenting was not improving despite intensive support and intervention, was sufficiently convincing for the Local Authority to move directly into care proceedings. This is significant because, as Hannon et al. (2010) demonstrated, children who are taken into care sooner rather than later in such cases fare better in the long term.
Changes in total areas of serious concern (scores of 4 or 5). All areas of the Action for Children Assessment Tool were implicated in practitioners' concern. Some cases showed only a single item, but it was not uncommon for five or six items to be identified to be a serious concern (scored at 4 or 5 on the scale). All three main areas of the tool showed obvious improvement for the whole population of cases (Fig. 2), so there were no aspects of neglect which could not be addressed successfully in most cases.

Discussion
The purpose of the study was to establish what worked in intervening with neglected children and to understand more about the factors that were linked to success. The issues considered below arose from the quantitative findings, illuminated by analysis of narrative data from the case files.

Home visits and the relationship between parent and worker
Among the factors that proved to be essential to the success of the initiative were the home visits and the relationship between parent and worker which had particular pertinence. Gaining access to children, and relating effectively to them and their parents and carers in their homes, is a deeply complex practice which Ferguson (2009) suggested is the core experience of "doing the work" in child protection but which is all too often ignored. Parental engagement is a key issue for services working with families where children may be at risk of abuse or other significant welfare problems (Platt, 2012), but for some parents, engagement can be highly problematic. It is not uncommon for these parents to refuse admission to professionals on home visits, or, once in the home, to prevent professionals relating directly to the child (Ferguson, 2009). The emerging outcomes from this study prompted a separate study to articulate the characteristics of effective relationships with service users. The findings (reported in terms of a skills framework) confirmed that maintaining a balance of support and challenge, being persistent in bringing parents to recognize their parenting shortcomings, being clear and direct so as to establish the appropriate scope of professional-parent relationships, demonstrating credibility and genuineness, presenting information sensitively and at an appropriate level, and adopting a solution-focused approach were the factors that brought about positive results (Crowther & Cowen, 2011).
It is possible that access was eased for the workers in this study through an understanding that they were not social workers. Indeed, the workers reported establishing effective relationships with parents and children, gaining trust and therefore access to challenge failings, offer guidance, and to provide reassurance and moral support. Nevertheless, it was vital in the first instance to establish an explicit, overt understanding with parents that the problem was neglect. This was often accomplished with parents who were hostile or clinically depressed, or who were otherwise deeply suspicious of agencies and their intentions. It was clear from letters and case conference minutes that workers gained access to houses and information which was sometimes not available to other workers and professionals. Indeed, they sometimes became the only workers with meaningful access and opportunity to observe the children and work with the parents. To achieve this access and to sustain an effective working relationship the practitioners had to demonstrate persistence and determination to engage parents, working through repeated rebuffs, and retaining the motivation to succeed.
Both the focus on home-visiting and the complexity of assessing, addressing, and evaluating neglect demanded a high degree of knowledge, skill, and experience of the practitioner. Complex circumstances, vulnerable families, and the breadth of factors to include in decision-making provide a challenge to any worker. That said, during the home visit the workers were able to identify (with parents) the wide range of problems that were stimulating or aggravating the neglect. They were then able to prioritize the problems which needed to be addressed first. Often, the pursuit of a single problem would involve a raft of activities, each of which might need to be taught and demonstrated, and then a sustained program of coaching, reminding, and motivating parents until a routine was established. Skill was needed to break this chaotic mass of difficulties down into manageable units of learning and then gradually to build the whole picture back up again into the full skills set required for adequate parenting.

The utility of the Action for Children Assessment Tool
The Action for Children Assessment Tool proved to be useful both in identifying key areas of neglect together with example descriptors and in promoting shared identification of failings and routes to improvement with parents. The tool also afforded the workers what Ferguson (2009) calls "mobilities" or the ability to move around and view the space in which children live. Its stable, objective nature and simple scoring system also engendered confidence in the assessment of problems and evaluation of outcomes. Using the instrument allowed parents to come to their own understanding of what was lacking in their parenting: a powerful encouragement to parents to make the necessary improvements and changes. The instrument was often completed in stages so that parents were presented with challenging but manageable targets. The whole complexity of the case was held by the worker while the parent was encouraged to focus on improving selected aspects of care. A revised version of the instrument now forms part of an Action for Children practitioners' toolkit. Care is taken to ensure that the instrument remains a supplement to practitioner judgment rather than a substitute for it (DfE, 2011).
Non-linear progression. Sequential assessment of progress using the Tool would often highlight the complexity of cases. Some total scores would increase as major risks were successfully addressed but additional, less urgent items came to the fore. Not uncommonly, as the worker gained the trust of parents, the parents would divulge the presence of additional problems. Alternatively, despite a trend of improvement in parenting and a reduction in scores, an untoward event could prompt a sudden and significant increase in concern. This might result from the introduction of another adult into the household or a crisis in relationship between parents. In these cases the worker would review the assessment, revise the plan and priorities, and demonstrate to parents how to react positively and effectively to the new problems. This might be viewed in one way as taking retrograde steps, yet each incident that was worked through exerted an incremental learning effect on parents. In this study, the best practice across all of the services was evidenced when the manager or lead practitioner engaged all the staff in reflecting on what the assessment tool was indicating and then supporting them to share the findings with confidence with more qualified colleagues.

Early age
Neglect can become an issue at any stage of a child's life (Horwath, 2007). The birth of an additional child, changes in parental relationships, the arrival of an adult who poses a risk, the onset of substance misuse, and many other factors might tip the balance and transform what was previously acceptable parenting into clearly neglectful parenting. Neglect may occur in early infancy for one child but in middle-childhood for an older sibling. There were cases in the evaluation in which older children were neglected as attention was focused on young siblings. However, the neurodevelopmental evidence is clear that for infants who are subject to neglect, rapid intervention is vital (HM Government, 2011;Howe, 2005). The age of children was not found to be an influence on outcomes for children in this study. In almost all cases in which neglect was so severe and improvement so inadequate as to require entry into care, all of the children became subject to care proceedings. The whole family situation was the key, with the focus mostly on parental response and ability such that the risk to all of the children was too great to be managed at home.

Early stage
The best results were clearly to be seen when intervention began early in the family's decline into neglectful parenting, highlighting that early intervention was vital. Early intervention might be taken to mean intervention with the smallest possible delay after neglect has been identified as a concern. The Action for Children Assessment Tool allowed for more certain identification of the presence of neglect and the specific elements which cause concern, notably in less clear-cut cases where practitioners' uncertainty might otherwise cause harmful delay. In this study, cases were split fairly evenly between those featuring entrenched neglect and others at an early stage of neglect. While a minority of the more severe cases were worked to successful transition to less intensive services, cases at earlier stages in the development of neglect proved especially susceptible to successful intervention and eventual return to universal services as used by most parents. While we were unable in this study to offer convincing evidence of how to distinguish reasonably quickly between cases in which intervention might be successful and those which would prove to be hopeless, we hope to pursue this in a further study.

Conclusion
The study demonstrated the ability to intervene successfully in most cases of neglect, even when neglect was a most serious concern. In cases where parents refused or were unable to respond positively, children move into care was expedited. The ability and willingness on the part of parents to engage with services was a crucial factor in deciding whether progress would be made or children taken into care. Further work is needed to investigate the factors in parents that support or militate against a positive response to offers of help for efficiency in the approach to borderline cases to be enhanced.
The work undertaken in outreach or home visiting by often unqualified support workers with especially difficult cases requires a high level of skill as well as experience. Further research is required to establish the means by which practitioners achieve this level of skill and expertise, and to isolate the constituent elements in order to inform training, mentorship and effective supervision.
The Action for Children Assessment Tool enabled practitioners to work with parents to establish a joint understanding of problematic aspects of parenting and to plan for staged improvements. It also provided a valuable source of evidence of objective assessment and review. Following review with practitioners, the instrument has since been developed further to reduce its size without damaging its effectiveness. Testing for continued effectiveness and retention of both objectivity and specificity will be necessary.