Teasing apart complex causal systems with novel statistical methods

Margarita Moreno-Betancur and John Carlin recently published a paper in the journal Epidemiology that provides critical insights on methods for causal mediation analysis, which is the statistical analysis of pathways to disease and other outcomes. This work identifies for the first time how mediation analysis can be used to emulate randomised trials evaluating interventions on mediating pathways, opening the way to estimating effects that would arise from intervening on mediating variables. Catch her video abstract here and the paper here

Previous work by Margarita in the same journal has the most viewed video abstract (watch it here). It proposed the first survival analysis model formally incorporating the possibility that death may be caused by multiple diseases acting together, by using all the causes of death mentioned on the death certificate (on average 3.3 causes/death in Australia!). This is critical with our aging population and the rise of chronic and degenerative diseases. See the paper here.