Bianca Lucia De Stavola

Introduction: This article summarises lessons learnt from the Health Outcomes for young People throughout Education (HOPE) Study and serves as a real world, transferable application for addressing causal questions using administrative data. The HOPE study applied causal methods to analyses of administrative data in Education and Child Health Insights from Linked Data (ECHILD) aimed at studying the effectiveness of provision for special educational needs and disability (SEND) on health and education outcomes. Methods: Defining causal questions regarding the impact of SEND provision required judicious mapping of the question onto the data, leading to the selection of appropriate measures of effect, transparent handling of the data and control of confounding factors to estimate effects. We adopted the target trial emulation framework to guide these steps. Having encountered specific computational challenges in estimating the effects of interest, we simulated data that resembled the HOPE study and used them to practice the implementation of alternative estimation methods and to study impact of some of their assumptions. Results: The creation and analysis of the simulated data provided valuable insights. First, we learned the importance of aligning the target of estimation with the causal question at hand. Second, we observed how deviations from assumptions specific to each estimation method can affect results. Third, we highlighted the benefits of employing alternative estimation methods as sensitivity tools that can aid the interpretation of the resulting estimates. Finally, we offer user-friendly code in two programming languages (R and Stata) and accompanying simulated data to facilitate the implementation of these methods for similar causal questions. Conclusion: We recommend users of administrative data to fully specify -and possibly revise- the causal questions they wish to address and to carefully examine and compare assumptions, implementation and results obtained using alternative estimation methods.

Background: Major congenital anomalies (CA) affect 2.3% of livebirths and are associated with lower educational attainment. Understanding attainment trajectories throughout primary school would inform parents, schools and organisations and help plan support. Objectives: We compared school enrolment and attainment at ages 5, 7 and 11 in children with different CA and their peers in England using the Education and Child Health Insights using Linked Data database. Methods: We included all singleton children born in NHS‐funded hospitals from September 2003 to August 2008 who enrolled in state‐funded schools at ages 4–5. CA were identified from hospital diagnoses, procedures or death records. We described school enrolment, school‐readiness, the percentage who sat curriculum assessments and who achieved expected English and Maths attainment at three ages. We estimated risk ratios of children with CA achieving expected attainment compared with peers, adjusting for sociodemographic factors. Results: Of 2,351,589 children enrolled at age 5, 78,847 (3.5%) had CA. At age 11, 88.7% of enrolled children with CA sat assessments versus 97.2% of peers. Proportionally fewer children with CA (45.7%) were school‐ready at age 5 versus peers (57.0%). For English, 56.9%, 55.4% and 65.3% of children with CA achieved expected levels at ages 5, 7 and 11 respectively, consistently 11%–12% fewer than peers; similar gaps persisted for Maths. Children with CA were less likely than peers to achieve expected attainment (adjusted risk ratio [aRR], 0.86, 95% confidence interval [CI] 0.85, 0.86), but this varied substantially (aRR 0.01, 95% CI 0.01, 0.02 for Down syndrome; aRR 1.04, 95% CI 0.96, 1.12 for unilateral renal agenesis). Conclusions: Attainment gaps between children with CA and peers remained unchanged across subjects and ages, with proportionally fewer sitting assessments at age 11. Better monitoring and support for these children from school entry could help optimise learning experiences and fulfil their academic potential.

IntroductionSchools worldwide balance whole-class teaching with additional provision for children with special educational needs or disability (SEND). Robust evidence on equity and effectiveness of SEND provision is essential to address growing demand and rising costs globally. ObjectivesTo synthesise findings from the Health Outcomes for young People throughout Education (HOPE) evaluation of variation in SEND provision and its impact on health and education outcomes in English primary schools. We integrated findings from 14 sub-studies using administrative data in the Education and Child Health Insights from Linked Data (ECHILD) database and 10 mixed methods sub-studies. MethodsAnalyses of ECHILD data followed children from birth to age 11 years. We examined how variation in SEND provision was associated with health conditions, and school, social and organisational factors. Using target trial emulation, we estimated the impact of SEND provision on hospital admissions, school absences and attainment. We surveyed and interviewed young people, parents, and professionals and reviewed information about services to understand SEND processes and contexts. ResultsOf 3.8 million children born 2004 to 2013, 30% had SEND provision recorded by age 11. Health conditions were only partially associated with SEND provision, which was also related to male gender, social disadvantage, low attainment and type of school. SEND provision modestly reduced rates of unauthorised absences in subgroups of children but showed no measurable benefit on hospital admissions or school attainment. Mixed methods studies highlighted benefits of early, responsive support, challenges posed by limited capacity, harms caused by delayed or inadequate provision, and need for parent advocacy to access SEND provision. DiscussionWeak evidence of benefits of SEND provision in causal analyses likely reflects unmeasured confounding, lack of measures of provision received and insensitive outcomes in ECHILD data. SEND policies need robust evidence from analyses across jurisdictions using administrative data, enhanced with better measures, experimental methods and contextual evaluation.

IntroductionAnalyses of population administrative data can often only be minimally adjusted due to a lack of control variables, potentially leading to bias due to residual confounding. ObjectivesWe aimed to use linked cohort data to help address residual confounding in analyses of population administrative data. One particular aim was to explore strategies for when linked cohort and population administrative data cannot be accessed together in the same environment (are 'siloed'). MethodsWe propose a multiple imputation-based approach, introduced through application to simulated data in three different scenarios related to the structure of the datasets. We then apply this approach to a real-world example -- examining the association between pupil mobility (changing schools at non-standard times) and Key Stage 2 (age 11) attainment using data from the UK National Pupil Database (NPD). The limited control variables available in the NPD are supplemented by multiple measures of socioeconomic deprivation captured in linked Millennium Cohort Study (MCS) data. ResultsIn our real-world example, we included 509,670 individuals in the population NPD data, of whom 7,768 (1.5\%) were MCS cohort members. The unadjusted estimate of -1.86 (95% CI -1.92, -1.81) for the association between pupil mobility and Key Stage 2 attainment was attenuated to -0.92 (95% CI -0.97, -0.88) through adjustment for the NPD control variables, and further attenuated to -0.76 (95% CI -0.86, -0.67) through adjustment for the MCS control variables. ConclusionsLinked cohort data can be used to address residual confounding in analyses of population administrative data, and our proposed approach performed well across a range of simulated and real-world scenarios. The underlying principles are widely applicable: any analysis of administrative data could potentially be strengthened by linking a subset of individuals into richer cohort data. More research is required to understand how these methods can be applied more broadly.

This study evaluated the effect of multi-level HIV prevention interventions delivered through DREAMS (Determined, Resilient, Empowered, AIDS-free, Motivated, and Safe) program on common mental disorders (CMD) among adolescents and young adults (AYA) in rural South Africa. We used longitudinal data from two cohorts of AYA aged 13–35 years (N = 2184 for cohort 1 and N = 3067 for cohort 2), recruited between May 2017 and December 2018, with data collected at baseline and follow-up surveys until 2019. Retention rates exceeded 75% in both cohorts. Multi-level HIV prevention interventions included individual-level services like HIV testing and counselling, and post-violence care, and community-level interventions such as social asset building, social protection, and school-based HIV education. Probable CMD was measured using a 14-item Shona Symptoms Questionnaire, with a score of ≥9 indicating CMD. We used inverse probability of treatment weighting adjusting for socio-demographics, food insecurity, experienced violence, sexual behaviour, HIV status and baseline CMD, to estimate the average effect of multi-level interventions on CMD, quantified as a risk difference (RD). Of the 3944 participants, 519 received no intervention, 1060 received community-level interventions, 814 received individual-level interventions and 1551 received multi-level interventions. The prevalence of CMD increased with age, reaching 27% among those aged <25 years. In the causal analysis, exposure to multi-level interventions relative to no intervention was estimated to increase CMD (RD = 4.59, 95% confidence interval (CI):1.61-7.78). Additionally, when comparing multi-level interventions to community-level interventions alone, community-level interventions were found to be protective against CMD among adolescent girls and young women in Cohort 1 (RD = 5.45, 95% CI:1.54-8.92), indicating a higher CMD risk in the multi-level intervention group. These findings suggest that multi-level HIV prevention interventions as currently delivered, may not meet mental health needs in vulnerable AYA. Mental health screening and care should be integrated into HIV services for early diagnosis and treatment of CMD.

Delays and inequalities in access to Special Educational Needs and Disability (SEND) provision in early childhood are likely to alter children’s social and education trajectories long-term. Effective policy solutions require robust evidence on the timing, duration, and contexts in which children with early indicators of need can access SEND provision. We conducted a longitudinal study of 983,652 singleton children born in England in 2006-08 using ECHILD, a database containing linked hospital and state-funded school records. Nearly 16% of the study population had recorded SEND provision in Year 1 (age 5/6 years; 14.1% SEN Support, provided to children with less severe difficulties; 1.7% Education and Health Care Plan (EHCP), intended for those with more severe difficulties). SEN Support was more prevalent among males, those with pre-existing chronic health conditions and characterised by disadvantage and low Early Year Foundation Stage profile (EYFSP) scores. EHCP followed similar patterns but associations with disadvantage were weaker. Voluntary and academy sponsor-led schools were less likely to provide EHCPs, once known imbalances in pupil intake were accounted for. Children entering state-funded education in nursery had higher prevalence of SEND provision throughout primary school than later entrants. EHCP increased gradually while SEN Support rose slowly, peaking at Year 2 (age 6/7 years) and then flattening. Children clustered into four trajectories of SEND provision from Year 1 to 6 labelled “ Never ”, “ Stable-High ”, “ Decreasing ” and “ Increasing”. Those in the “ Never ” trajectory (77%) tended to live in income advantaged areas, had lower levels of chronic health conditions and higher EYFSP scores than peers. Children in the “ Increasing ” (i.e. delayed) trajectory were more likely to attend academy sponsor-led schools. Results suggest that there are inequalities in type and timing of SEND provision that are associated with school governance.

Background Evidence is lacking on whether Special Education Needs (SEN) provision for children with cerebral palsy improves their school attainment and reduces their hospital utilisation compared to unexposed peers. Methods We assessed the causal impact of SEN provision (None, SEN Support – arranged by the school-, and Educational and Healthcare Plan [EHCP] – arranged by the local government) at start of compulsory schooling (Year 1; age 5/6 years) on unplanned hospital utilisation, school absences during primary school and mathematics attainment at Key Stage 1 (KS1, taken at age 6/7) and 2 (KS2, taken at 10/11). We used a target trial emulation framework to define groups with similar needs from linked hospital and education data on children born 2003 to 2013 in England with cerebral palsy recorded in hospital before Year 1. We employed propensity score-based methods to control for the non-random allocation to SEN provision in this population. Results In our study population, 45% (1483/3275) of children affected by cerebral palsy had SEN Support and 42% had an EHCP in Year 1. Comparing scenarios where SEN Support was/was not provided in Year 1, we found some indication that SEN provision was associated with higher rates of unplanned hospital admissions, similar rates of total absences and lower rates of unauthorised absences. SEN provision was also associated with lower probabilities of achieving the expected mathematics level at KS1 but similar probabilities at KS2. Comparing EHCP to SEN Support, we found evidence of increased rates for unplanned hospitalization, an indication of decreased rates for unauthorised absences, and a decreased probability of achieving the expected KS2 level. Sensitivity analyses that modified the definition of the cerebral palsy population confirmed these results, except for rates of unplanned hospital admissions that were increased when comparing SEN provision versus none. Discussion We found some weak and conflicting evidence on the impact of SEN provision on health and education outcomes of children affected by cerebral palsy. Although there was possibly an indication of increased rates of unplanned hospital admissions, there was also a possibly positive effect on rates of unauthorised absences. Given that the rates of total absences did not appear to be affected, it is however possible that SEN provision has an impact on the coding of absences more that on its rate. With regards to mathematical achievement, again results are mixed, indicating some levelling up by the end of primary school (when KS2 levels are assessed). For all results we cannot discount however the impact of likely residual confounding bias.

Background There is limited quantitative evidence on how equitable special educational needs and disability (SEND) provision is, particularly among children with complex health conditions. We compared the first recorded SEND provision in state-funded nurseries and primary schools (ages 2 to 11 years) for children with cerebral palsy, by gender and indicators of socioeconomic circumstances. Methods We used linked individual-level state-funded hospital and school records from the Education and Child Health Insights from Linked Data (ECHILD) database to create a cohort of children with cerebral palsy born in England between September 2003 and August 2012. We used inverse probability weighted pooled logistic regression to assess whether there were inequities (i.e., differences that are unfair or avoidable) in the first recorded SEND provision, stratified by school Year of entry into state-funded education. Comparisons were expressed in terms of cumulative probabilities. We separately assessed the first recorded education, health and care plans (EHCPs), which are the higher tier of SEND provision. Results Of the 5,670 children in the cohort, 5,399 (95.2%) had any SEND provision and 4,098 (72.3%) had an EHCP recorded during follow up. Male gender and free school meal eligibility were associated with higher estimated cumulative probabilities of any SEND provision during primary school, whereas children living in the most deprived areas had lower cumulative probabilities of EHCPs. Conclusions Our findings suggest that current SEND practices are inequitable for children with cerebral palsy.

Background: Major congenital anomalies (CAs) occur in 2.3% of livebirths and are associated with lower educational attainment in affected children. Understanding how attainment changes throughout primary school would inform parents, schools and organisations and help plan support. Objectives: We compared school enrolment and attainment at ages 5, 7 and 11 in children with different CAs and their peers in England using linked administrative hospital and education data in the ECHILD database. Methods: We included all singleton children born in NHS-funded hospitals from September 2003 to August 2008 who enrolled in state-funded schools at age 4-5. CAs were identified from hospital diagnoses, procedures or death records. We described school enrolment, school-readiness, the percentages of children who sat curriculum assessments and who achieved expected levels in English and Maths at three ages. We estimated risk ratios of children with CAs achieving expected levels compared with peers, adjusting for sociodemographic factors. Results: Of 2,351,589 singleton children enrolled at age 5, 78,847 (3.5%) had CAs. At age 11, 88.7% of enrolled children with CAs sat assessments versus 97.2% of peers. Proportionally fewer children with CAs (45.7%) were school-ready at age 5 versus peers (57.0%). For English, 56.9%, 55.4% and 65.3% of children with CAs achieved expected levels at ages 5, 7 and 11 respectively, consistently 11%-12% fewer than peers; similar gaps persisted for Maths. Children with CAs were on average less likely than peers to achieve expected levels [adjusted risk ratio, aRR (95%CI): 0.86 (0.85,0.86)] but this varied substantially across CA subgroups [aRR (95%CI) range: 0.01 (0.01,0.02) to 1.04 (0.96,1.12)]. Conclusion: The attainment gap between children with CAs and peers remained unchanged across subjects and ages, with proportionally fewer sitting assessments at age 11. Better monitoring and support for these children from school entry could help optimise learning experiences and fulfil their academic potential. SYNOPSIS: Study question: What are the patterns of educational attainment in children with major congenital anomalies (CAs) throughout primary school? What is already known: Studies of regional or registry-specific cohorts of children with CAs showed that proportionally fewer achieve expected attainment relative to peers at single stages of national assessments. There is limited evidence on children’s participation in assessments and how attainment gaps change throughout primary school at a population level. What the study adds: Longitudinal analysis of whole-population cohorts from ages 4 to 11 found that attainment gaps between children with and without CAs remained largely constant across ages. Over half of children with CAs were assessed as not ’school-ready’ at age 5. Whilst almost all children with CAs remained enrolled in school at age 11, one in nine did not participate in assessments.

IntroductionAnalyses of population administrative data can often only be minimally adjusted due to a lack of control variables, potentially leading to bias due to residual confounding.ObjectivesWe aimed to use linked cohort data to help address residual confounding in analyses of population administrative data. One particular aim was to explore strategies for when linked cohort and population administrative data cannot be accessed together in the same environment (are “siloed”). MethodsWe propose a multiple imputation-based approach, introduced through application to simulated data in three different scenarios related to the structure of the datasets. We then apply this approach to a real-world example – examining the association between pupil mobility (changing schools at non-standard times) and Key Stage 2 (age 11) attainment using data from the UK National Pupil Database (NPD). The limited control variables available in the NPD are supplemented by multiple measures of socioeconomic deprivation captured in linked Millennium Cohort Study (MCS) data.ResultsIn our real-world example, we included 509,670 individuals in the population NPD data, of whom 7,768 (1.5%) were MCS cohort members. The unadjusted estimate of -1.86 (95% CI -1.92, -1.81) for the association between pupil mobility and Key Stage 2 attainment was attenuated to -0.92 (95% CI -0.97, -0.88) through adjustment for the NPD control variables, and further attenuated to -0.76 (95% CI -0.86, -0.67) through adjustment for the MCS control variables. ConclusionsLinked cohort data can be used to address residual confounding in analyses of population administrative data, and our proposed approach performed well across a range of simulated and real-world scenarios. The underlying principles are widely applicable: any analysis of administrative data could potentially be strengthened by linking a subset of individuals into richer cohort data. More research is required to understand how these methods can be applied more broadly.