With robust standard errors, actually sensible in binary outcomes, IMO.

I haven't thought much about it for survival type outcomes, but curious to see what others think!

Reporting observations correctly is a critical job of a statistician! I've been blocked for saying this!!

Do you think whatever "estimate" you produce in an observational study has adjusted for all known and unknown confounders? Congrats, your adjusted estimate is confounded too

Who cares if it's confounded? Just report what was observed

Your parameters are probably poorly measured, poorly parametrized, and have all sorts of informative missingness (perhaps you can tell the kind of stats I do from this)

At least I know there's some sort of weird truth behind the possibly confounded observed data. Who knows what kind of insanity lies in whatever your modeled data says

How do we know it is clearly confounded?! And why should I trust your poorly defined model more than WHAT WAS ACTUALLY OBSERVED?!

It tells you WHAT ACTUALLY HAPPENED. Not whatever wild made up assumptions you made in your model, which I am suspicious of. Why on earth would I trust your model

To not report them is hiding information. Am I missing someone here?