Event

G-formula is a popular approach for estimating treatment or exposure effects from longitudinal data that are subject to time-varying confounding. G-formula estimation is typically performed by Monte-Carlo simulation, where potential outcomes under the treatment regimes of interest are simulated from models fitted to the original dataset. Inference for the resulting estimates is usually performed using bootstrapping.

In practice the dataset to be analysed often has missing values, which are commonly handled using the method of multiple imputation (MI). I will talk about how Bayesian multiple imputation methods for creating synthetic datasets can be used to perform G-formula for causal inference, and that a by-product of this is that we can use multiple imputation to impute both missing values in the original data and missing counterfactual values for the treatment regimes of interest.

I will describe the approach and how it differs from regular multiple imputation for missing data, illustrating its performance in simulations and application using data from cystic fibrosis.

Jonathan Bartlett is a Professor in Medical Statistics at the London School of Hygiene & Tropical Medicine. His research interests are focused around missing data and causal inference methods, and more recently, how these can be applied to target different estimands in clinical trials. He has held previous positions at AstraZeneca and the University of Bath, and maintains a blog thestatsgeek.com.

Date and time

Thursday 24 July, 4-5pm AEST

Join

This seminar will be held over Zoom and it will be recorded.

Please click this URL to join: https://monash.zoom.us/j/83722878167?pwd=StJNUUorBW07CVPbiFMbyRbBQAfaA0.1
    Or, go to https://monash.zoom.us/join and enter meeting ID: 837 2287 8167 and passcode: 495815