Linking data to improve outcomes for neurodivergent young people

This blog by ADR UK Research Fellow Dr Justin Yang explores the risks neurodivergent pupils face in secondary school, and how linked data can help identify opportunities for earlier, more effective support.

Each year, numerous neurodivergent young people (who might be autistic, have ADHD, or face other learning differences), enter an important stage of life. Exam stress rises, friends shift, and support systems struggle to keep up.

The numbers are stark: neurodivergent pupils are more likely to miss lessons due to poor health, to self-harm, or to be formally excluded or taught outside mainstream classrooms. Each missed lesson, hospital visit, or exclusion hearing is an opportunity to identify need. However, we’ve lacked the joined-up evidence to understanding why and how these potential warning lights flash for some pupils and not others – until now.

What I will examine

My fellowship will look at four important questions to help understand the experiences of neurodivergent pupils in Key Stage 4:

• How often do neurodivergent pupils in Key Stage 4 face absences for medical reasons or show hospital-presenting signs of self-harm or suicide? Are there personal or clinical factors that heighten or lessen risk?
• Which neurodivergent pupils end up excluded or moved to alternative provision, and what health-related events precede these decisions?
• Do different levels of special educational needs support - ranging from mainstream schooling with extra help, to specialist units, to separate schools - change the odds of positive outcomes among neurodivergent pupils?
• To what extent does the “postcode lottery” exist for adverse events, and what might that tell us about local services and differences?

How I’m doing it

To answer these questions and others, I’ll use data from the Education and Child Health Insights from Linked Data (ECHILD) dataset for all pupils in Key Stage 4 from 2015-2020. This anonymised, highly secure dataset links every state pupil’s school records to their hospital records.

For the first time, using this linked data, we can relate how a hospital visit on a Monday might connect to a long absence on Tuesday or the rest of the week, or how switching to a special school might affect self-harm a year later.

The study will run in three stages:

1. Mapping patterns: I will count and compare pupils with different types of neurodivergent special educational needs to understand who misses long stretches of school, presents to hospital for self-harm, or gets excluded. I’ll also look at how these patterns emerge, and how they differ by group characteristics, such as gender or region.
2. Understanding causes: Using an approach called “target trial emulation,” I’ll try to mimic a randomised experiment within the data. For instance, I will compare otherwise similar autistic pupils starting Key Stage 4 in a mainstream school with those in a special school, to see which group records fewer admissions to hospital for self-harm. This approach helps us understand potential causes rather than just associations.
3. Exploring geography: I’ll map health-related harms and exclusions across England’s local and unitary authorities to understand if certain areas show patterns of higher or lower risks, accounting for potential differences. By doing so, I’ll highlight the characteristics of places with higher risk, as well as those with lower risk.

Who is this for?

Numbers and maps alone won’t change lives, so people with lived experience and professional experience will be a part of this project.

I’ve already met with an advisory group of neurodivergent young people who completed Key Stage 4 in England from 2015 onwards. Together, we tackled tough questions such as the limitations of the data in representing the full picture (especially for pupils who effectively “mask”). But we also explored the benefits of this research.

I’ll also be meeting with educators, special educational needs professionals, and other people who support neurodivergent pupils, gaining their perspectives into this research and how it might maximise benefits and impact.

What difference will this make?

By spring 2026, this project will deliver:

• Open access code and guidance so other researchers can use ECHILD more effectively
• Evidence briefs for policymakers responsible for revising the special educational needs and disabilities (SEND) system, highlighting key areas for early support
• Interactive maps for local authorities to identify ways to improve SEND provision
• Accessible multimedia outputs highlighting key findings, which will be written in plain language.

My hope is simple: when a young person starts to struggle, schools and health teams should have hard evidence on what works, rather than relying on trial and error. Linked data gives us that evidence. By connecting the dots between education and health, we can spot gaps sooner, act smarter, and build a fairer future for every child who thinks and learns differently.

Explore more ADR UK blogs and news posts to keep up to date with the power of administrative data.