When code has a bug in it, that is way more of an author problem than a reviewer problem. I don’t think it’s right to expect reviewers to find bugs (or to rely on it). It’s way harder to do that unless you actually run the code.

So, I would suggest that to fix the problem of buggy code making it into production, you do more things at the author level, since the author should be running the code a lot through tests and in the debugger. If you have a code reviews checklist, give it to the author, who should be reviewing their own code before PR’ing it.

Then, the reviewer should be verifying that there is evidence that the author actually did thorough work. The review checklist should be verifying the author provided evidence that the author did the pre-PR checklist. Check the tests, check their coverage, check UI screenshots, check that code is doing “something” with edgecases and that those are tested. Instead of looking for bugs, the reviewer should be learning the code and looking for higher level issues.

The author should be providing code that is correct and easy to see that it’s correct. If the reviewer doesn’t think it’s obviously correct, they should reject the PR. The presence of such a review will make the author do more thorough work.