Phenotyping neurodisability in hospital records in England: a national birth cohort using linked administrative data

Background: Children with neurodisability often have complex healthcare and educational needs. Evidence from linked administrative health and education data could improve joint working between services.

Objective: To develop a diagnostic code list to identify neurodisability in hospital admission records; to assess the representativeness of this phenotype by characterising children with hospital-recorded neurodisability and their outcomes.

Methods: We developed a national cohort of singletons born in England between 2003 and 2009, including a nested cohort of children enrolled in primary school, using linked health and education data from the Education and Child Health Insights from Linked Data (ECHILD) database. With expert clinicians, we developed an algorithm based on diagnostic information from hospital records to phenotype children with hospital-recorded neurodisability. We described rates of mortality, planned/unplanned admissions up to 11 years old, and school-recorded special educational needs (SEN) provision, as proxy measures of the complexity of a child's needs, overall and for over 40 neurodisability subgroups.

Results: Of 3,580,225 children in the birth cohort, 3.6% had hospital-recorded neurodisability by age 11. The most frequent subgroups included developmental disorders, autism, epilepsy, perinatal brain injury, and cerebral palsy. Children with hospital-recorded neurodisability had higher mortality and planned/unplanned admission rates compared with their peers, and they accounted for 26% of all planned and 14% of all unplanned hospital admissions before age 11. The nested primary school cohort included 2,956,299 pupils (82.6% of all births), 3.7% of whom had hospital-recorded neurodisability. 75% of children with hospital-recorded neurodisability had any school-recorded SEN provision, and 39% had a record of more intensive provision (compared to 30% and 2.4%, respectively, for their peers).

Conclusions: We derived a phenotype for hospital-recorded neurodisability, which affects 1 in 28 primary school children in England, with high rates of hospital admissions and SEN provision. This phenotype and its subgroups can be used by service providers and researchers to examine inequalities and inform resource and service provision.