Developing, validating, and implementating innovative statistical approaches in causal inference

Randomized clinical trials provide the highest level of evidence of the effect of an intervention. By eliminating confounding bias and ensuring a balanced distribution of potential confounding factors, randomization ensures a fair comparison between groups of individuals. However, randomized clinical trials are increasingly expensive, time-consuming, and can sometimes pose ethical problems. Cohort data collected through routine visits are a ready-made source of information. Under certain assumptions, advanced statistical methods known as 'G-methods' are able to correct for selection bias present in observational data and produce results similar to randomized trials. We aim to contribute to the development, validation, and implementation of ‘G methods’.