Keeping the Promise - health and wellbeing of care experienced children and young people: longitudinal quantitative evidence review

Results

Routine administrative data

Children Looked After Statistics (CLAS)

CLAS is the Scottish Government’s annual administrative collection which gathers data from the 32 local authorities to provide a comprehensive overview of all formally looked after children in Scotland. CLAS compiles information on children’s characteristics, placements, and the legal grounds under which children are placed in care. According to the latest CLAS 2023-24 publication, 10% of looked after children had a recorded disability. For another 25% of children, their disability status was unknown or not recorded.

CLAS is the foundation for the Looked After Children Longitudinal Dataset which currently covers a timeframe of 2008 to 2019. This dataset combines individual returns and as a consequence provides a fuller picture of children’s care experience. It is designed to be used by researchers when a care history is required. This will be updated in 2025 using more recent data up to 2023/24. This dataset is highly relevant as it captures all care experienced individuals under the legal care of their local authority before and after the introduction of The Promise. It also presents linkage opportunities with other datasets. The search revealed several research projects that used the CLAS dataset to explore outcomes in care experienced children and young people. These are discussed in turn below.

CLAS linked datasets

Children’s Health in Care in Scotland (CHiCS)

The CHiCS research programme, led by the University of Glasgow, offers a powerful approach to understanding health outcomes among care experienced children and young people. CHiCS links a wide range of routinely collected datasets held by government departments, local authorities, and the NHS spanning social work, education, health, and vital events. These include CLAS, hospital inpatient and day case episodes, psychiatric admissions, accident and emergency (A&E) attendances, prescribing, and mortality records. A key strength of CHiCS is its longitudinal cohort design. The study tracks over 13,000 children and young people from the 2009-10 CLAS return, covering those born between 1990 and 2004, with a follow-up extending to 2016 (when participants were aged 11 to 27). This can support analysis of long-term health trajectories across childhood, adolescence, and early adulthood.

In contrast to other data sources (discussed later), which often rely on self-report and are prone to sampling bias or small care experienced subsamples, CHiCS uses administrative records. What sets CHiCS apart from other data sources reviewed is its ability to directly identify care experienced individuals through formal looked after status recorded in CLAS data. This avoids reliance on proxy indicators of care experience, such as household structure or type of caregiver. It also allows for analysis by placement type (e.g., foster care, residential care, at-home supervision). This level of granularity can provide a clearer, more nuanced understanding of the associations between different care environments and health outcomes. Moreover, CHiCS enables robust longitudinal analysis of a range of outcomes, as well as service use and mortality risk, with several peer-reviewed publications emerging from the dataset.

Descriptive analysis of the CHiCS dataset by Allik and colleagues (2021) revealed that care experienced children had higher rates of health events recorded than children in the general population. This included higher mortality rates (5.5 times higher), mental health inpatient and day cases (5.2 times higher), maternity hospitalisations (4.3 times higher), and A&E attendances (2.1 times higher). The care experienced cohort also had higher rates of mental health service use and hospital admissions compared to children in the general population. Nearly one in five (19.7%) of the care experienced cohort had been prescribed medication for depression. This was more than double the rate observed in children in the general population (8.1%). Similarly, outpatient psychiatric clinic attendance was recorded for 20.1% of the care experienced group. This compared to just 5.5% among children in the general population. Acute inpatient admissions for mental and behavioural disorders were also disproportionately higher in the care experienced group (2.2%) relative to children in the general population (0.3%). Hospitalisations due to injuries, drug poisoning, and other external causes were more prevalent among care experienced children. For example, drug poisoning admissions affected 6.1% of this group – over six times higher than 0.9% in children in the general population. These figures underscore the heightened vulnerability of care experienced children and young people to a range of mental health and risk-related harms.

Using the CHiCS cohort, Brown and colleagues (2021)found that all-cause mortality was three times as high in the care experienced cohort than in those who were not care-experienced (after adjusting for confounding factors: age group, sex, deprivation and disability). The risk of death from external causes, such as suicide and drug deaths, was almost six times higher in the care experienced group. This study highlights the long-term consequences of childhood adversity and being care experienced. Although reasons for entering care vary, many children enter care because of neglect, abuse, parental substance use, or exposure to domestic violence. These factors are recognised as Adverse Childhood Experiences (ACEs; Scottish Government, 2025), and known to be associated with poorer mental health, substance misuse, and early death. Thus, care experience should be understood not only as a marker of social vulnerability, but also as a proxy for cumulative childhood trauma.

Allik and colleagues (2024) compared risk periods before, during, and after care of the care experienced CHiCS cohort. Results showed that care experienced children and young people were not significantly more likely to be diagnosed with asthma or diabetes but had higher rates of epilepsy than the general population cohort. However, for all three conditions, hospitalisation rates were markedly higher among the care experienced group, especially before entering and after leaving care. Interestingly, risk of hospitalisation was lower during periods of care placement. The authors suggest that entering formal care may offer temporary health stability or improved access to primary care. However, the higher hospitalisation outside care episodes reinforces the need for earlier intervention and sustained support during transitions out of care.

Gedeon and colleagues (2023) explored the association between chronic conditions (epilepsy, asthma, and diabetes) and mental health hospitalisation in the CHiCS cohort. They found that care experienced children (5.1%) were significantly more likely to have a mental health hospitalisation than children in the general population (0.8%). In the general population, all three chronic conditions were associated with an increased risk of mental health hospitalisation, with epilepsy showing the highest risk. However, among the care experienced group only asthma was statistically associated with increased mental health hospitalisation risk (epilepsy and diabetes were not, potentially due to smaller sample sizes). The authors suggest that care experienced children face a high baseline risk of mental health hospitalisation, regardless of chronic condition.

CHiCS Summary

Taken together, the CHiCS studies show that care experienced children and young people face poorer health outcomes than non-care experienced children. While formal care placements may offer temporary stability, health risks remain elevated both before and after care. Although CHiCS offers a uniquely detailed and robust dataset, its findings are based on data up to 2016. Therefore, it cannot provide real-time insights into the more recent landscape since the implementation of The Promise. Furthermore, as it is based on the CLAS dataset, CHiCS also excludes children who were only looked after in the preschool period, as they do not receive a SCN, limiting linkage for this group.

The value of CHiCS could be maximised by extending follow-up of the original cohort, enabling analysis of outcomes for care experienced individuals now aged 21–35. This could support analysis of long-term health trajectories, late-emerging conditions, mortality, and service use. While this cohort predates The Promise, it provides a robust baseline for assessing whether systemic inequalities persist and may capture early signs of change during its initial implementation. Encouragingly, according to a summary report by the CHiCS research team there are plans to extend the study’s follow-up period to the end of 2022 (including the national lockdowns). This includes plans to broaden its scope to cover areas such as childhood vaccinations, maternal health, health behaviours before birth, and birth outcomes. Thus, it would be of interest to consider any planned CHiCS extensions, which would provide deeper insight into health outcomes in adulthood.

One potential approach could be the development of a news CHiCS-style cohort, focused on children entering care more recently in light of The Promise implementation. This would enable direct evaluation of outcomes during and after the introduction of The Promise. Applying a similar CHiCS linkage methodology would enable comparability with earlier data. Stratification by pre- and post-2020 entry into care could help identify early impacts and allow subgroup analyses by placement type, local authority, and length of time in care. Although initial follow-up would limit adult outcome measurement, such a resource could underpin long-term monitoring.

However, there are various practical challenges associated with data linkage projects such as CHiCS. It is unclear whether the additional benefits would justify the investment given ongoing developments on the Children’s Health and Monitoring The Promise PHS project. Nonetheless, CHiCS remains an invaluable resource for identifying persistent health inequalities. It also helps us understand the long-term consequences of care experience, particularly for benchmarking historical challenges.

Other research

The search revealed other research using the CLAS dataset to explore long term health and wellbeing outcomes in care experienced children and young people. The Permanently Progressing? study tracked children who entered care aged five or under in 2012-13. The three-phased study follows this cohort and assesses outcomes during key life stages (early childhood, middle childhood, and into late adolescence/early adulthood). Phase 1 (2014–2018) reported on a wide range of outcomes of 433 children (aged three to nine years old) who became looked after away from home during 2012-13 and remained (or were again) looked after away from home one year later. Phase 2 (2020-24) built on findings of Phase 1 by revisiting the same cohort to explore outcomes during middle childhood (when children were between nine and 16 years old). Both Phase 1 (N=166) and Phase 2 (N=98) included analysis of data from children’s caregiver surveys.

Data collected from caregivers at Phase 1 found that around a third (31%) reported children as either having, or currently being assessed for, long-standing illnesses, disabilities or health conditions. For context, the authors compared this to findings from a study by Bradshaw and colleagues (2015), where 14% of a representative sample of three-year-olds had a long-standing health issue. By Phase 2, the rate of children reported as having a diagnosis or undergoing assessment for any long-standing illnesses, disabilities, or health conditions had more than doubled (68%). This was compared to findings from a Scottish Government (2022) publication where 15% of a sample of 14 year olds had a long-standing health issue.

Caregivers also completed the Strengths and Difficulties Questionnaire (SDQ), a validated screening tool for emotional and behavioural difficulties. Results indicated that 28% of the children had scores suggestive of likely emotional and behavioural difficulties. This was a rate two to three times higher than the 10% typically found in general population samples. By Phase 2, 50% of children had SDQ scores indicative of likely emotional and behavioural concerns. This was markedly higher than during Phase 1. In addition, caregivers completed the Relationship Problems Questionnaire (RPQ), which measures potential relationship and attachment difficulties. Overall, 25% of children had scores indicating a likelihood of relationship and attachment problems. This increased to 29% by Phase 2. Despite these challenges, nearly all caregivers described children’s physical health as either ‘healthy’ or ‘very healthy’ at both Phase 1 (99%) and Phase 2 (98%). However, at Phase 1 children looked after away from home were significantly less likely to be described as 'very healthy' (67%), compared to those in kinship care (82%) or on an adoption pathway (85%). These findings indicate that care experienced children face higher risks of long-standing health conditions, emotional and behavioural difficulties, and relationship and attachment problems. The findings also underscore the importance of looking at children’s outcomes by placement type, as outcomes may vary depending on the care environment.

McMahon and colleagues (2018) examined dental health outcomes and access to dental services among looked after children. The study used data from the 2012 Scottish pupil census and linked this to a range of routinely collected datasets. This included CLAS, the National Dental Inspection Programme, the Management Information and Dental Accounting System, and hospital discharge records (SMR01). Findings revealed that rates of regular dental attendance were lower for looked after children (51%) compared to the general population (63%). Hospital dental extractions under general anaesthesia were higher among looked after children (9% compared 5% of the general population). Nearly a quarter (23%) of looked after children were assessed as needing urgent dental treatment at age five. This was more than double the rate observed among non-looked after children (10%). These disparities persisted even after adjustment for deprivation, suggesting that care experience confers additional oral health risk beyond socioeconomic background. This study provided evidence that care experienced children face barriers to preventive oral health services, despite Scotland’s universal healthcare system. The findings imply need for more integrated and proactive health assessments for children in care. They also highlight the need for targeted support to reduce inequalities in dental health outcomes. A limitation of this study is that children who were only looked after in the preschool period could not be included, as linkage relied on identifiers assigned at school entry.

Fleming and colleagues (2021) compared health outcomes of looked after children (looked after anytime between 2009 and 2012) with their general population peers. Results, based on a mean follow up of four years, showed that children who experienced care had higher rates of special educational needs. They were also more likely to be treated for epilepsy, attention deficit hyperactivity disorder (ADHD), depression, and neurodevelopmental multimorbidity. Overall, care experienced children were over 1.2 times more likely to be hospitalised for any reason. For hospital admissions due to injuries, care experienced children were almost twice as likely to be admitted. Most notably, they were over five times more likely to be hospitalised because of self-harm than their general population peers. These elevated risks remained even after adjusting for socioeconomic and maternity factors, suggesting that being in care likely contributes to children’s mental health vulnerabilities. The association between care status and physical health conditions was more mixed. Care experienced children had a moderately elevated risk of epilepsy. No significant differences were found in the treatment rates for diabetes, asthma, or skin disorders. This suggests that the health burden associated with care experienced is not generalised across all conditions but likely concentrated in areas linked to ACEs.

Soraghan and Porter (2024) analysed data of children looked after in kinship care in Scotland between 2008 and 2019 and linked this to health visitor records. Findings revealed that at the 27–30-month health review, 37% (135 out of 365) of children with experience of kinship care had at least one developmental concern. This was notably higher than the 18-19% of developmental concerns observed in the general population. Assessments from visits at age 4-5 years showed that 31% of children in kinship care (33 out of 108) had at least one developmental concern, higher than 10-13% observed in the general population. Emotional and behavioural difficulties were the most frequently reported concerns at both developmental stages. These findings suggest that children in kinship care are more likely to present with early developmental concerns than their peers. The study also highlights the value of linking health and care data to identify unmet needs early in the life course. Finally, it underscores the importance of ensuring consistent access to developmental assessments for children in care. A limitation of this study is that children who were only looked after before age five could not be included due to linkage constraints.

Early Child Development (ECD) Statistics

ECD statistics is an annual official statistics release by PHS. It reports on universal Child Health Reviews at three key stages: 13-15 months, 27-30 months, and 4-5 years. These reviews are delivered primarily by Health Visitors and include screening for developmental concerns across eight domains: speech, language and communication; gross motor; fine motor; problem solving; personal/social; emotional/behavioural; vision; and hearing. These reviews include an indicator for whether a child is looked after at the time of assessment, providing a valuable source of information on early developmental outcomes for this group. As outlined in the ECD technical report, the ‘looked after’ indicator is recorded by Health Visitors within the Child Health Reviews and has high completeness (98%, 98% and 96% at the 13-15 month, 27-30 month and 4-5 year reviews respectively in 2023/24). Accuracy has been validated against the Scottish Government CLAS dataset, with figures aligning within expected ranges since 2016/17. PHS consider data recording of looked after status to be of adequate quality to be informative about the developmental outcomes of this population from 2016/17.

At the 27–30-month review, 40.1% of looked after children had at least one developmental concern recorded in 2023/24, compared with 16.6% of non-looked after children. This reflects a gap of 23.5 percentage points. In 2018/19 (before the introduction of The Promise) this gap was narrower at 14.1 percentage points (28.5% vs. 14.4%). At the 4-5 year review, 43.5% of looked after children had a developmental concern recorded in 2023/24 compared with 18.2% of non-looked after children. This reflects a gap of 25.3 percentage points. This gap was similar in 2018/19 (37.2% vs. 11.9%, also 25.3 percentage points). The most common concerns at both stages related to speech, language and communication, emotional and behavioural development, and personal/social development. These findings show that developmental concerns are consistently much higher among looked after children. The gap widened for the 27-30 month review since 2018/19, while the gap at the 4-5 year review remained unchanged, even though prevalence rose for both groups.

This is a valuable source of data about younger care experienced children, where there is less data currently available. However, this data cannot be used in isolation to assess the effectiveness of The Promise specifically without considering wider context and additional indicators. Timing of entry into care, recording practices, changes in population characteristics, or service access could all affect these numbers independently of The Promise. Future ECD statistics could make greater use of the placement type information already recorded within child health reviews (as outlined in the ECD technical report), and provide breakdowns by individual characteristics. This would enable more granular analysis of how different care settings and demographic factors are associated with developmental outcomes. Such detail could help identify patterns within subgroups, highlight areas where targeted interventions may be needed, and provide richer context for interpreting trends over time.

Scotland’s Census

The Census offers a near complete population coverage, providing a comprehensive snapshot of demographic and social circumstances at a national scale. It also includes useful data on children and young people living in residential establishments, which are often underrepresented in other household surveys (discussed later on). Finally, the Census has potential for linkage with other datasets.

The Census includes a household questionnaire and a communal establishments questionnaire. It also includes and individual questionnaire completed for each household member and each communal establishment resident. The individual questionnaire provides detailed information on respondents’ disabilities and self-assessed general health. This offers greater detail than the standalone CLAS dataset which is limited to overall percentages of disability (i.e. percentages recorded as Yes/ No/ Not yet assessed/ Not known/recorded). Therefore, the Census may provide valuable insight into children’s health and disability profiles. This includes around a decade before the introduction of The Promise (Census 2011) and during the very early implementation years of the Promise (Census 2022). The next Census 2031 could provide context after The Promise 24-30 Plan concludes.

However, identifying care experienced individuals in the Census is complex. The Census communal establishments questionnaire should technically capture all children looked after living in residential care. One category that could be analysed in more detail in the communal establishment questionnaire is ‘Children’s home’. Figures for this category are not readily available on the Census table builder but could be requested from NRS. Nonetheless, this classification is still broad and may encompass a wide spectrum of residential care settings. It does not distinguish between establishments housing looked after residents from those housed for other reasons. Similarly, other available categories (e.g., ‘education’, ‘hostel or shelter’, ‘youth hostel’) likely include residents beyond just those who were looked after by their local authority. The absence of a direct indicator for care experience in the Census risks misidentification or undercounting of this population.

Children and young people looked after in the community (e.g., those in kinship care, foster care, or under supervision at home) should all be captured by the Census household questionnaire. However, the absence of a dedicated care indicator in the Census makes identifying this population challenging. The recently published Kinship Care Statistics analysis addressed this by using a proxy indicator from the Census. The proxy indicator estimated the number of children living in kinship care with relatives, covering both formal and informal arrangements. This proxy defined kinship care as children not living with their parents and recorded as having one of the following relationships to the household reference person: stepchild, sibling (including step- or half-siblings), grandchild, or other relation. Children living with non-relatives, such as family friends, who may also meet the definition of kinship care were excluded from this estimate. Capturing these cases would require using the Census ‘Unrelated’ category, which likely encompasses a wide range of circumstances, including foster placements and non-care-related living arrangements. Furthermore, there is no way to identify children living at home with their parents but under a compulsory supervision order (see Annex 2) who, by definition, are also care experienced. As a result, there is a risk of misclassification and underestimation of the care experienced population. This makes it difficult to produce accurate comparisons between care experienced children and non-care experienced children.

The Kinship Care Statistics report explored the characteristics of children in kinship care and compared them with those referred to as ‘those in other households’. Notably, the breakdowns by disability indicate that the proportion of children in kinship care with a disability was almost double that of children in other households (15% compared with 8%). This analysis could be extended by incorporating more detailed categories of disability status and drawing on self-reported health data from the Census. However, the absence of a direct care-experience indicator limits the precision of such comparisons. Without it, there is a risk that the ‘other households’ group includes care experienced children, which may blur distinctions between groups and complicate interpretation of any observed differences.

The search did not identify any other research using Census data to explore health and wellbeing outcomes in care experienced individuals. The Scottish Longitudinal Study (SLS) leverages Census data (from 1991 onward) as part of its data linkage approach to other data types (e.g. vital events, health, education). However, there was no record of SLS projects specifically on care experienced children and young people.

To make the Census a more useful source for analysing outcomes for care experienced children, the most important improvement would be to introduce a direct indicator of care experience status. This would enable clearer identification across residential and community settings. Refining communal establishment categories to distinguish care-specific settings would also help reduce misclassification.

Birth cohort studies and longitudinal studies in Scotland and the UK

Understanding Society Survey (USS)

The USS is a longitudinal UK study collecting extensive data across health, wellbeing, family circumstances, and socioeconomic factors. However, several limitations affect its utility for researching Scotland’s care experienced population. Children in foster care placements are likely classified as Temporary Sample Members in USS, which means they are not consistently followed up over time, unlike Original Sample Members. This poses challenges for longitudinal analyses due to potential gaps in observations across waves. Up to wave 14, USS data focuses primarily on foster care placements. While the survey captures information about children living with relatives other than their parents, it does not explicitly categorize these arrangements as kinship care. Thus, deriving precise figures on kinship care placements is challenging. Questions introduced from wave 15 onwards on guardianship relationships could in future help infer some kinship arrangements. However, these cannot definitively establish legal care status.

Furthermore, USS does not distinguish between children living at home with their biological parents under compulsory supervision orders and those who are not. This poses challenges for comparing outcomes between an estimated care experienced cohort and children in the general population. Additionally, Borkowska (2019) found that up to 66% of USS’s care experienced sample through 2017 were based in England (145 out of 219 unique observations). This suggests that the Scottish subsample is likely very small. This means USS has limited generalisability to the Scottish care experienced population as this subgroup is not well covered. Indeed, a search of published literature did not identify any studies in scope. Lastly, like other household surveys, USS relies on sampling of households meaning that children and young people living in residential care are omitted. Therefore, it is unlikely that USS offers useful means for evaluating outcomes of care experienced children and young people in Scotland. To make USS more useful, improvements could include introducing explicit questions on care status and whether a child is under local authority supervision while living at home with their parents. Expanding the Scottish sample size through targeted oversampling could also enhance representativeness. However, residential care populations would nonetheless remain outside USS’s scope.

Growing Up in Scotland (GUS)

GUS is a longitudinal, Scotland-specific birth cohort study initiated in 2005. It was designed to track the lives of children and families across multiple domains including health, wellbeing, education, and service use. However, GUS does not include a direct indicator for whether a child is care experienced. Care experience could potentially be estimated by indicators in GUS such as reason for child’s move, living arrangements, and parental responsibility. However, this is unlikely to be an accurate identification of this group. Additionally, the number of children who are care experienced in GUS is likely very small. This limits the robustness of statistical analyses, especially when comparing subgroups. Another challenge is that children with more complex or unstable care experiences may be harder to follow up over multiple waves. While GUS makes substantial efforts to retain participants, attrition may be higher among care experienced children, particularly those not living with their birth parents.

The search did not identify any research using GUS data to explore health and wellbeing outcomes in the care experienced population. Given the likely small and unrepresentative sample of care experienced children and young people within GUS, its usefulness for this purpose is very limited.

Millennium Cohort Study (MCS)

The Millennium Cohort Study (MCS) is a UK-wide longitudinal study following children born in 2000-2001, including a Scottish subsample. The MCS collects rich data on children’s health, education, family circumstances, and social development across multiple waves. However, it does not include a variable that captures care status. Identifying care status would require reliance on indirect proxies, such as household composition, which cannot reliably distinguish formal care placements from other private living arrangements. Furthermore, the Scottish subsample is small (around 12%). The number of children who may have had care experience is likely even more limited for robust subgroup analysis. Attrition bias is a particular concern: children who experience care are more likely to drop out of longitudinal studies due to placement changes, instability, or disengagement. This could potentially further reduce sample representativeness. The search did not identify any published research that uses MCS data to analyse outcomes of care experienced children and young people in Scotland. Without complex and potentially unreliable inference strategies, MCS cannot currently contribute meaningful evidence to this review.

British Cohort Study 1970 (BCS70)

The BCS70 follows individuals born in 1970 across the UK, including Scotland, collecting extensive data on health, education, and family life. Care experience in BCS70 is identified using a specific indicator based on parental reports at ages five, 10, and 16, where parents were asked whether the child had ever been ‘in out-of-home care.’ This includes placements outside the family home, such as foster care or residential care, although it does not distinguish kinship care or supervision-at-home arrangements. The search did not identify any analyses specifically using BCS70 to study care experienced individuals in Scotland. However, one study by Cameron and colleagues (2018) examined long-term outcomes of care experienced individuals in the overall British cohort. Data from childhood sweeps and retrospective reports indicate that 616 individuals (3.8% of the study population) had been in care before age 17. Outcome data at age 30 were available for around 350 of these care experienced participants. Based on this, it is plausible that the Scottish sample size within the cohort is even smaller. Lastly, Given the cohort’s age and major shifts in care systems since the 1970s, BCS70 has limited relevance for understanding the health and wellbeing of today’s care experienced young people..

National Child Development Study 1958 (NCDS)

The NCDS follows individuals born in one week of 1958 across Great Britain and includes long-term data on life outcomes. While it includes Scottish participants, NCDS does not include any direct questions on care experience. Some early life circumstances such as time in institutional care may be captured through retrospective reporting. However, these are not sufficient to reliably identify care experienced individuals by today’s definitions. No studies were found using the NCDS to investigate care experienced populations in Scotland. Due to its historical cohort and decades of policy change, the NCDS cannot meaningfully inform current policy development or evaluation of The Promise.