Citation

Hartwig FP, Davey Smith G. Int. J. Epidemiol. 2024; 53(4): dyae094.

Copyright

(Copyright © 2024, International Epidemiological Association, Publisher Oxford University Press)

DOI

10.1093/ije/dyae094

PMID

39067478

Abstract

Instrumental variable (IV) analysis is a research design that allows valid causal conclusions about the relationship between an exposure and an outcome even if there is unmeasured confounding or reverse causation, which would bias conventional observational associations. This analysis requires that a third variable (the IV) satisfies three conditions:

Additional assumptions are needed for a precise interpretation of the effect estimate from an IV analysis.2

Since IV analysis also requires assumptions, its results can only be trusted to the extent to which the underlying assumptions are deemed plausible. Therefore, a crucial aspect of an IV study is to provide theoretical and empirical justification for the assumptions. Although it is impossible to guarantee that assumptions (ii) and (iii) hold, it is possible to assess their plausibility empirically using rather simple strategies.3-6 Even if there is good theoretical justification for the validity of an IV, we believe that empirical evidence should also be provided whenever possible to corroborate (or not) the theory and therefore provide a more robust justification for the validity of the analysis. ...

Language: en

Keywords

Humans; Mental Health; *Mental Disorders/epidemiology; *Transportation; Bicycling/statistics & numerical data; Walking