In more ways than one. it is not only about residual confounding from missing out on covariates. It is also about the fact that different combinations of confounders can swing the outcome in opposite directions. It's called vibration of effects (VoE) and plagues roughly two-thirds of exposure variables, alcohol being one of them.