Abstract: Observing treatment on a single occasion does not reflect real life clinical management where treatment may vary over time. Patients may start, change or stop treatment repeatedly over time, with each treatment choice being influenced by time-varying and invariant covariates measured up to that time point, If these time varying covariates also predict the outcome and are influenced by preceding treatment decisions, they are termed time-dependent confounders. This seminar will focus on a few select approaches of accounting for time-dependent confounders in assessment of treatment effectiveness.

Speaker: Dr Milica Blagoejvic-Bucknall, a lecturer in Statistics based at Keele University with interest and expertise in observational longitudinal data analysis, in particular electronic health record data.

---