Target trial emulation for policy evaluation

Speakers: Bianca de Stavola and Lorraine Dearden

There are many challenges in designing studies to address policy evaluation using linked administrative data. Designing such studies highlights the difficulties of defining the target population (e.g., children in need of Special Educational Needs (SEN) provision, a very heterogeneous group), qualifying the exposure (e.g., school-recorded SEN), understanding its drivers whilst also accounting for clustering of provision by local authority and school. Additionally, data completeness and temporal variability in coding practices raise issues of how to account for selection bias (due to selection into the study) and residual confounding (due to poorly measured confounders). In this seminar, we will discuss how adopting a target trial emulation framework in this context helps clarify the limitations of the data and the questions that can be addressed.

Watch the seminar recording