Apostolos Gkatzionis Seminar

AUEB STATISTICS SEMINAR SERIES FEBRUARY 2021

Apostolos Gkatzionis, MRC Integrative Epidemiology Unit, University of Bristol

Selection bias in Mendelian randomization

FRIDAY 19/2/2021, 13:30

ABSTRACT

Mendelian randomization (MR) is an approach for causal inference in epidemiological studies. MR uses genetic data as instrumental variables in order to assess the causal effect of a risk factor on an outcome of interest. The approach has gained in popularity over the last few years, due to the availability of large-scale genetics datasets and the development of relevant statistical methods.

This presentation is split into three parts. In the first part, we provide a brief introduction to Mendelian randomization. The second part concerns selection bias in MR studies. We describe how selection bias can invalidate the key instrumental variable assumptions and explore by simulation how severe the bias is likely to be in practice. In the third part, we present some ongoing work on using an instrumental variable for selection in order to adjust for selection bias, in either observational studies or MR analyses.

(Presentation slides can be found here)