An introduction to instrumental variable assumptions, validation and estimation

Abstract

The instrumental variable method has been employed within economics to infer causality in the presence of unmeasured confounding. Emphasising the parallels to randomisation may increase understanding of the underlying assumptions within epidemiology. An instrument is a variable that predicts exposure, but conditional on exposure shows no independent association with the outcome. The random assignment in trials is an example of what would be expected to be an ideal instrument, but instruments can also be found in observational settings with a naturally varying phenomenon e.g. geographical variation, physical distance to facility or physician’s preference. The fourth identifying assumption has received less attention, but is essential for the generalisability of estimated effects. The instrument identifies the group of compliers in which exposure is pseudo-randomly assigned leading to exchangeability with regard to unmeasured confounders. Underlying assumptions can only partially be tested empirically and require subject-matter knowledge. Future studies employing instruments should carefully seek to validate all four assumptions, possibly drawing on parallels to randomisation.

Impact of opioid dose escalation on the development of substance use disorders, accidents, self‐inflicted injuries, opioid overdoses and alcohol and non‐opioid drug‐related overdoses: a retrospective cohort study

“…Given that none of the approaches outlined above can account for potential unobserved confounding, instrumental variable (IV) models were estimated . The use of IV analyses hinges on finding a valid IV which is a variable that influences exposure—in this case, being more likely to escalate opioid dose—but does not directly influence the outcome (e.g.…”

Two-step Randomisation: Applying the Results of Small Feasibility Studies of Interventions to Large-scale Mendelian Randomisation Studies to Robustly Infer Causal Effects on Clinical Endpoints

“…We cannot therefore discard other potential effects of the interventions which are not observed in the feasibility trial due to low power, that we may otherwise observe in a definitive, adequately powered trial (2). Also important in this case is the 4 th IV assumption, where the observed effect of the intervention or treatment is a local average, that is, it may only apply to a specific subgroup of patients, and it may incorrectly translate at individual level or in a different subgroup, an issue that affects both RCTs and MR studies and may be seen in some clinical settings (21)(22)(23)(24).…”

Smoking is a cause of amyotrophic lateral sclerosis. High low‐density lipoprotein cholesterol levels? Unsure

“…Mendelian randomization (MR) methods follow this reasoning. They apply instrumental variable (IV) methods developed in the field of economics to infer causality in the presence of unmeasured confounding . The concepts underlying IV or MR methods are illustrated in the Figure, and the underlying assumptions are listed in the Supplementary Table.…”