Abstract

Observational epidemiological studies suffer from many potential biases, from confounding and from reverse causation, and this limits their ability to robustly identify causal associations. In other observational sciences, notably econometrics, the use of instrumental variable approaches has been one approach to strengthening causal inferences in non-experimental situations. The use of germline genetic variants as instruments for modifiable (non-genetic) risk factors is one form of instrumental variables analysis that can be implemented within observational epidemiological studies. The method has been referred to as “Mendelian randomisation”, and can be considered as analogous to randomised controlled trials. This presentation will briefly define Mendelian randomisation and instrumental variable analysis; demonstrate how the instrumental variable approach differs from multivariable regression (the more common approach to dealing with confounding in observational analysis) and discuss the potential and limitations of genetic variants as instrumental variables. Recent methodological developments that might be used to address some of the limitations will also be discussed.