Mendel’s laws, Mendelian randomization and causal inference in observational data: substantive and nomenclatural issues

Summary

This response paper defends Mendelian randomization (MR) as a valid causal inference method in epidemiology against criticisms that it receives excessive attention and should be renamed. The authors demonstrate that substantive methodological concerns—including population stratification, confounding, weak instrument bias, and pleiotropy—have been systematically addressed within the MR field. Notably, they argue that MR shows greater methodological rigour and self-examination than conventional observational epidemiology, which has historically recycled failed approaches into new research areas.

UK applicability

As a methodological paper focused on epidemiological inference techniques, UK applicability depends on whether UK epidemiologists and public health researchers adopt MR approaches in their studies. The paper's argument for MR's methodological superiority could inform UK health research standards and evidence hierarchies used in policy-making.

Key measures

Methodological robustness of Mendelian randomization relative to conventional observational epidemiology; examination of handling of population stratification, confounding, weak instrument bias, and pleiotropy

Outcomes reported

The paper evaluates the methodological validity and advancement of Mendelian randomization (MR) as a causal inference tool in observational epidemiology, addressing criticisms regarding population stratification, confounding, instrument bias, and pleiotropy. It demonstrates how MR has confronted and addressed these methodological challenges compared to conventional observational epidemiology.

Topic tags