In ROBINS-I, a confounding factor is a variable that affects both the intervention a person receives and the outcome being studied, which can make the observed association differ from the true causal effect. In other words, it can make one treatment look better or worse than it really is.

Simple example

If older patients are more likely to receive a certain treatment and are also more likely to have worse outcomes, age is a confounder. The study might then wrongly attribute the worse outcome to the treatment instead of to age.

Why it matters in ROBINS-I

ROBINS-I uses confounding as one of its main bias domains because non- randomized studies do not balance groups the way randomization does. The tool specifically notes baseline confounding and, in some cases, time-varying confounding when factors change after the study starts.

Practical way to think about it

A good confounder usually has three features:

  1. It is related to the intervention received.
  2. It is related to the outcome.
  3. It is not caused by the intervention itself.

That is why ROBINS-I asks reviewers to identify prognostic factors that may influence treatment choice and outcomes.

In one sentence

A confounding factor in ROBINS-I is any outside variable that distorts the estimated effect of an intervention because it influences both treatment assignment and outcome.

Would you like a very short ROBINS-I example you can use in a paper or assignment?