All Times EDT
Keywords: Self-control method, causal effect, target maximum likelihood estimation
There has been an increasing interest in causal inference using real-world data (RWD) for regulatory and healthcare decision-making. One common objective of such practice is to provide additional insights for approved drugs in reporting post-marketing safety. This paper extends the causal inference approach for case-control studies by Rose and van der Laan (2009) to self-controlled case series (SCCS) studies. First introduced in 1995, the SCCS method uses cases as their own controls in which all time-invariant confounders are automatically controlled, rendering the possibility of the causality assessment for time-varying effects that can be efficiently carried out by using targeted maximum likelihood estimation (TMLE). The proposed approach is applied to a real-world dataset to investigate the causal relationships between an immunotherapy and a rare AE observed post therapy.