Without accounting for Special Education Needs (SEN) requirements, prior evidence demonstrates SEN provision increases school absences. Our objective was to estimate the impact of SEN provision in Year 1 (age 5/6 years) on unauthorised absences during primary education in a population with SEN needs, those with Cerebral Palsy.

Using linked hospital and school data from England, we accounted for group imbalances by defining a cohort of children born between 2003 and 2013 with recorded Cerebral Palsy, excluding those with other major congenital anomalies. Using target trial emulation, we defined the target population aligning exposure and follow-up at Year 1. Using health and education data, we assessed the propensity overlap for three groups of exposure: No Provision, “SEN Support” and “Statemented SEN”. We estimated the average treatment effect (ATE) of these interventions on cumulative unauthorised absence rates by the end of primary education (age 10/11) using inverse probability weighting.

We found 4659 children, 769 with No Provision, 2172 with SEN Support and 1718 with Statemented SEN. We could not compare No Provision versus Statemented SEN due of lack of propensity overlap. When comparing SEN Support in reference to No Provision, we found a similar rate of unauthorised absences in the two groups with a marginal incident rate ratio (IRR) of 0.91 [95% Confidence Interval (CI): 0.54 – 1.36]. When comparing Statemented SEN in reference to SEN provision, the former led to a reduction in unauthorised absences by the end of primary education, IRR 0.80 [95%CI: 0.66 – 0.98]. We confirmed our results using regression adjustment.

Through administrative data linkage, we created a well-defined cohort which was expected to have similar need for SEN provision. Through use of a target trial emulation framework, we reduced bias when creating the study population, and demonstrated that certain SEN categories would comparatively reduce unauthorised absence rates using IPW.