Data Resource Profile: Children Looked After Return (CLA)

Early exposure to adversity, such as abuse or neglect, is associated with poorer outcomes across social, education and health domains. Children in care (referred to as looked-after children in the UK) are a vulnerable group who experience adversity serious enough for the state to intervene in family life and place them under the supervision of child protection services within the home or, more frequently, to remove the child and place them in out-ofhome care (OHC). In England, placement in OHC can be voluntary (i.e. with parental consent) or mandated by a court. Some looked-after children have complex health needs and are voluntarily placed in temporary care in order to provide respite to their parents, but the majority of children in OHC are removed from their parents for reasons related to abuse or neglect. Being in OHC is an indicator of serious childhood adversity and a predictor of future adverse health, education and social outcomes. For example, children in OHC have poorer mental and physical health than their peers, are more likely to engage in risky behaviours such as smoking, drinking and drug-taking and have higher rates of teenage pregnancy and premature death. The causes of these adverse outcomes are complex and there is considerable heterogeneity among looked-after children. Some variation in outcomes has been associated with key characteristics of the care children receive while being looked after (e.g. age at first entry, setting, duration, stability) or with their exit from the social care system (e.g. destination, re-entry). For example, children in foster care have better mental health outcomes than those in residential group care, and psychiatric disorders are more common among children who experience multiple placement moves. It is therefore important to determine the prevalence among the child population of being placed in OHC and to explore how different types or patterns of care are associated with outcomes, both in childhood and in later life. Many studies of looked-after children in the UK are based on surveys; however, these may have selection and/or recall biases, and an alternative administrative data source that can be used is the Children Looked After Return (CLA). The CLA offers an important resource to improve understanding about the characteristics of children placed in OHC, how patterns of care vary across the country and are changing over time, and the relationships between the type or pattern of care and subsequent outcomes.


Childhood adversity
Early exposure to adversity, such as abuse or neglect, is associated with poorer outcomes across social, education and health domains. 1,2 Children in care (referred to as looked-after children in the UK 3 ) are a vulnerable group who experience adversity serious enough for the state to intervene in family life and place them under the supervision of child protection services within the home or, more frequently, to remove the child and place them in out-ofhome care (OHC). 4 In England, placement in OHC can be voluntary (i.e. with parental consent) or mandated by a court. Some looked-after children have complex health needs and are voluntarily placed in temporary care in order to provide respite to their parents, 5 but the majority of children in OHC are removed from their parents for reasons related to abuse or neglect. 6 Being in OHC is an indicator of serious childhood adversity and a predictor of future adverse health, education and social outcomes. 7 For example, children in OHC have poorer mental and physical health than their peers, [8][9][10] are more likely to engage in risky behaviours such as smoking, drinking and drug-taking 11 and have higher rates of teenage pregnancy and premature death. 12 The causes of these adverse outcomes are complex and there is considerable heterogeneity among looked-after children. [13][14][15] Some variation in outcomes has been associated with key characteristics of the care children receive while being looked after (e.g. age at first entry, setting, duration, stability) or with their exit from the social care system (e.g. destination, re-entry). [16][17][18] For example, children in foster care have better mental health outcomes than those in residential group care, 11 and psychiatric disorders are more common among children who experience multiple placement moves. 10 It is therefore important to determine the prevalence among the child population of being placed in OHC and to explore how different types or patterns of care are associated with outcomes, both in childhood and in later life.
Many studies of looked-after children in the UK are based on surveys; 8-10 however, these may have selection and/or recall biases, and an alternative administrative data source that can be used is the Children Looked After Return (CLA). The CLA offers an important resource to improve understanding about the characteristics of children placed in OHC, how patterns of care vary across the country and are changing over time, and the relationships between the type or pattern of care and subsequent outcomes.

Purpose and scope of the Children Looked After Return
In England, children's social care services are delivered at local government level (i.e. by local authorities). The CLA is a national individual-level dataset held by the Department for Education (DfE), which contains information on all looked-after children and recent care leavers in England. Data collection began in 1992 (Table 1) and is ongoing via an annual online census of local authorities. Initially, data collection was mandated for all children in England who were looked after in the year ending 31 March 1992; however, between 1998 and 2003 it was restricted to a one-third sample (selected as children with a day of birth divisible by three) before reverting to include all looked-after children in 2004. The CLA contains detailed care histories for looked-after children including the start and end dates of each episode of OHC.
According to the DfE, the purpose of the CLA is to monitor the care and outcomes of looked-after children (while in care and on reaching adulthood) and to enable evaluation of the potential effects of government policy initiatives. 5 Outcome data were first collected in 1999, but were limited to the activity of children in care at age 16 years (i.e. taking exams, in further education, working, etc.). Since 2009, the outcome data collected by local authorities have been expanded to include child-level information on health-related outcomes such as immunizations, health checks and Strengths & Difficulties Questionnaire (SDQ) scores. However, outcome data are only collected for children in continuous care for 12 months or more. In 2002, collection of data on the activity and accommodation of care leavers at age 19 began with further follow-up at other ages introduced in later years.
All looked-after children are included in CLA (with the exception of the previously-described sample restrictions between 1998 and 2003). However, the CLA does not include private fostering arrangements in which a child is cared for by an adult who is not a close relative (i.e. someone other than a parent, grandparent, sibling, aunt or uncle). 19 The most recent CLA for the year ending 31 March 2015 contained details of 99 230 looked-after children-the highest figure since 1985. 20 Coverage of care leavers in CLA is not complete; information was collected for 84% (n ¼ 22 510) in 2014 and 88% (n ¼ 23 170) in 2015. 20

Dataset production
Each year, all 150 local authorities in England must submit details of the looked-after children in their area and the   What was the child doing on their birthday (i.e. in education, employment)?
• What type of accommodation was the child living in on their birthday?
• Was the accommodation suitable?
The years in which these variables were collected in the CLA vary and are described in full in official Department for Education guidance. The underlined variables are available for request from the Department for Education through the National Pupil Dataset team. Other variables are not routinely available to researchers, but can be requested. *Care episodes recorded in CLA are funded by the state via local authorities, but may be delivered on their behalf through approved private organizations (e.g. a looked-after child may be placed with an agency foster carer or in a children's home run by a charity).
care provided to them during the period 1 April to 31 March to the DfE, via an online census. The number of children in care varies from year to year, but in the most recent period of data collection (1 April 2014 to 31 March 2015) data were collected for 99 230 children. Local authorities must also provide information for specific groups of care leavers (i.e. young people who were looked after as adolescents and whose 19th to 21st birthday occurred during the preceding year). Data must be returned and checked by local authorities before the end of June. 5 A national dataset is then collated by DfE. Aggregate tables and summary statistics (at national and local authority levels) are then produced by DfE and published online. 21 Dataset structure: episodes and periods of care In the CLA, a child's care record is divided into episodes. An episode is the length of time a child is looked after under the same legal status and in the same placement. When a child's legal status and/or placement changes, a new episode begins. 5 The start and end date of each individual episode is recorded in the CLA, and an episode cannot be less than 24 h. Episodes of care can be in the home (under supervision) or in alternative out-of-home accommodation (e.g. with a foster carer or in a children's home, Supplementary

Measures collected in CLA
The measures collected in CLA have changed over time but can be broadly grouped as child characteristics, episode information, and indicators and outcomes of care (Table 2).

Child characteristics
When a child becomes looked after by a local authority for the first time, they are assigned a child ID-the main identifier in the CLA. This allows a child's record of care to be linked over time and enables longitudinal analyses. The demographic information collected in the CLA is limited to date of birth, gender and ethnicity (18 categories). Names are not collected. Whether a child is an unaccompanied asylum seeker (or a mother, for girls who are looked after) is also recorded, but this information is not routinely available to researchers. A pseudonymized unique pupil number (UPN) is recorded for looked-after children who attend a maintained (or state-funded) school or nursery in England, 22 which allows linkage of CLA data to other education and social care datasets held by DfE.

Episode information
Detailed information related to each episode of care is collected in the CLA; for example, start and end dates, placement type, location and provider. Placement type describes the setting in which a child is cared for. Children may be placed at home with their parents while being looked after, but the majority are removed and placed in OHC. 6 OHC placements include foster care by relatives, friends, strangers or potential adopters; group care in children's homes, residential schools, care homes or residential units; independent living in a bed and breakfast (B&B), flat or bedsit and 'other' settings such as young offender institutes and prisons. The codes used to record placement type have changed over time and are described in Supplementary  Table 1. When a child's placement changes (even to another placement of the same type) a new episode begins. However, only placements lasting 24 h or more are recorded; therefore if multiple placement changes occur in 1 day, only the final placement is recorded. 5 The reason a child becomes looked after is recorded in the CLA as their 'category of need'. These hierarchical categories are: abuse or neglect, child's disability, parental illness or disability, family in acute stress, family dysfunction, socially unacceptable behaviour, low income and absent parenting. Though it is likely that a child will become looked after for a combination of the above reasons, only one (the highest ordered in the list) is recorded. 23 Before the 1 April 2000, a more detailed variable was used to capture the reason a child was looked after, and the relationship between these former 'reasons looked after' and current 'categories of need' is described in Supplementary Table 2.
The legal status of a child describes the legal framework under which a child enters the social care system. For example, child protection is used to ensure the safety of a child who is considered to be in need, and this legal status category includes emergency protection orders and police protection powers (used in urgent cases where rapid intervention is required) and child assessment orders (used in non-emergency cases where there are suspicions but no convincing evidence of actual or likely harm). 24 Children can also be looked after voluntarily (i.e. with parental consent) under Section 20 of the Children Act 1989. 3 Though it is possible for a child to have multiple legal statuses (e.g. to be under a care order and awaiting trial), only the most recent legal status is recorded in the CLA. As for placement setting, if there are multiple changes in 1 day, only the final legal status is recorded. 5 The codes used to record legal status have changed over time and are described in Supplementary  Table 3 (available as Supplementary data at IJE online).
When a child ceases to be looked after, the reason the period of care ended is recorded. For example, a child may cease to be looked after because they return home to their parents or are adopted. They may also leave care through the granting of a residence or special guardianship order which confers differing levels of parental responsibility to a guardian 25 (such as a relative or former foster carer). The codes used to record the reason OHC ceases and how a child exits the social care system are described in Supplementary Table 4 (available as Supplementary data at IJE online).

Indicators and outcomes of care
One of purposes of the CLA is to monitor outcomes of looked-after children while in care and on reaching adulthood; however, outcomes are generally recorded only for children who have been in continuous care for 12 months or more. The data recorded for these children in long-term care include whether they were convicted of a crime, identified as having a substance misuse problem, offered intervention to treat substance misuse, and had up-to-date health checks, dental examinations and immunizations. Children aged 4 to 16 years should also have an annual SDQ score recorded (which can be used as an indicator of emotional or behavioural disorders). Similarly, outcome data are only collected for 'relevant and eligible' care leavers as defined by current DfE guidance, 5 i.e. a young person who was looked after at the age of 16 or 17 and had been looked after for at least 13 weeks after the age of 14. The outcomes recorded for care leavers are participation in education and/or employment and living arrangements, currently at age 19 to 21. Indicators of care, such as time to adoption, participation in statutory case reviews and being missing from care, are also collected in CLA. Data related to indicators and outcomes of care are not routinely available to researchers but can be requested.

Data quality checks
The CLA undergoes a number of automated validation checks when data are being returning by local authorities; 26 for example, fields that are blank or contain an invalid value will be flagged for review and correction. Unlikely/impossible sequences of dates or combinations of legal status and placement are also automatically flagged, as is information that contradicts records entered in previous years for the same child. During the validation checks, local authorities may correct errors or update previous years' data (i.e. enter an end date for an episode of care that had been ongoing at the time of the latest census).

Describing trends
DfE statistical tables and reports are published annually and are readily available to the general public online [https:// www.gov.uk/government/organisations/department-for-edu cation/about/statistics]. These DfE tables include information on the rate of looked-after children in England by local authority, which can be used as an area-level indicator of childhood adversity. CLA data are used to monitor the use of OHC in England and outcomes of looked-after children over time. For example, recent DfE reports indicate that the rates of substance misuse and offending are falling among children in care, 20 but the proportion of care leavers not in education, employment or training (NEET) has increased. 27

Monitoring outcomes using linkage
Only limited educational outcomes are recorded in CLA; however, another DfE administrative dataset [(The National Pupil Dataset (NPD)] contains detailed information on a broad range of educational outcomes, including absences, exclusions, Special Educational Needs (SEN) support and type of school attended. Since 2006, NPD and CLA data have been routinely linked via UPN 28 by the DfE; this linked dataset has been used to describe the relative educational outcomes for children looked after in continuous care for 12 months or more. 27 Pseudonymized linked CLA-NPD data have also been used by researchers. For example, a recent study by Sebba et al. 29 explored the effects of the type and timing of OHC on children's educational outcomes, specifically the attainment of children eligible to sit GCSE exams in 2013. This study also involved linkage to a third DfE dataset that contains details of children who are referred to social services but are not placed in care (the Children In Need (CIN) dataset), available from 2009. This additional linkage allowed researchers to conduct more granular analyses in terms of exposure to adversity during childhood. Children in care (due to serious adversity) were compared with children in need (who experienced adversity at a level that was insufficient to warrant state involvement) and with all other children who were not in care or in need. The results of these analyses suggest that some of the gap in educational attainment between children in care and their peers can be attributed to differences in deprivation and SDQ scores, but early placement in long-term foster care can have a protective effect on attainment. 29 However, this analysis was limited to children in care at the time of sitting their GCSE exams (at age 15/16), and early exposure to care was simplistically defined as being in continuous care for at least 12 months and having also been in care during late primary school years (Key Stage 2).

Analyses across the child life course
DfE reports make limited use of the rich longitudinal records of care available in CLA. For example, DfE annual reports focus on the number of placement moves a child has during a year, rather than the total number of moves they experience during their total time in care. However, the CLA can also be used to generate evidence on the child life course of care. For example, in a recent study that used longitudinal CLA data, we calculated the proportion of children in England who ever entered care throughout childhood, using synthetic birth cohorts. We found that one in 30 children born 1992-94 (3.3%) had entered OHC by age 18 years, 30 a much higher figure than the 0.6-0.9% of children who spend time in care in any given year as reported by the DfE. 20 The cumulative proportion of children ever entering care also appeared to be increasing (particularly among infants) and was disproportionately higher among Black, Mixed or Other ethnicity children. Decomposition of these changes over time vis-a-vis concurrent changes in the ethnic composition of the child population indicated that the overall increase in the proportion of children entering care was primarily due to an increase in the proportion of White children entering care, rather than increased ethnic diversity among children in England.

Cross-national comparisons
Aggregate or child-level CLA data can also be used to explore variation in child protection and social care systems between different countries. For example, Gilbert et al.

Strengths
The main strengths of the CLA are that it has national coverage and is an administrative dataset, thereby negating issues of recall or selection bias associated with surveybased studies of OHC that rely on self-report by care leavers or caregivers. The CLA has collected cross-sectional data annually since 1992, and this allows for changes over time in the population of looked-after children, and the characteristics of the care they receive, to be reliably described. The CLA is also a longitudinal dataset that contains complete histories for children and allows care trajectories to be explored in detail. An additional strength is that summary statistics are freely available to download online for use as an indicator of local authority-level adversity during childhood. Furthermore, the DfE recognizes the unique value of the CLA as a longitudinal data source for policy evaluation and research and they are committed to continued data collection and improvement of content. For example, permanence is a central component of current social care policy in England; 33 therefore, indicators of a breakdown in a permanent exit from care (i.e. adoption, special guardianship order or residence order) and of permanence within social care system via long-term foster care, were recently introduced. 23 CLA data can also be used to evaluate local policies as data are available broken down by local authority.

Weaknesses
The restriction of data collection between 1998 and 2003 limits the power of the longitudinal dataset, particularly when exploring variation by local authority or for relatively rare placements or outcomes (e.g. death). A further limitation is that child ID is a local authority-specific identifier. If a child is looked after in more than one local authority, they will be assigned multiple child IDs, consequently preventing linkage of care records across these administrative boundaries. Similarly, when a child is adopted they receive a new legal identity. Therefore, if they subsequently become looked after again, they are assigned a new child ID. This means that a child's records of care pre-and post-adoption are not linked. However, the main limitation of the CLA is that (as it is an administrative dataset) it does not contain baseline characteristics of children entering care or their families or provide detailed information related to the care and support looked-after children receive (e.g., interventions provided, parental contact, placement with siblings, etc.). Furthermore, outcome data are only collected for specific groups of looked-after children and care leavers, and linkage to other non-DfE datasets (related to health or justice, for example) is not facilitated as names are not collected.

Data resource access
Aggregate statistical tables, annual reports and documentation related to CLA are available to the public at [https:// www.gov.uk/search?q ¼ childrenþlookedþafterþ]. Requests for pseudonymized child-level CLA can be made by researchers through the NPD team at DfE. Data related to child characteristics and episodes of care (underlined in Figure 2) are routinely available for request from 2006 onwards. Other years of data or variables (such as SDQ score, postcode or UPN) are not routinely available, but can be requested and have been supplied to researchers in the past. 29 Though CLA data is pseudonymized, it is considered 'tier 1' (i.e. sensitive personal information); therefore to obtain an extract, researchers must complete an information security questionnaire and application form, which are considered by an advisory panel. When making an application for CLA data, the need for each requested variable must be clearly justified by researchers. Applications can also be made to link CLA data to NPD and/or CIN data. Further application details and documents are available at [https:// www.gov.uk/guidance/national-pupil-database-apply-for-adata-extract].

Supplementary Data
Supplementary data are available at IJE online.