There's a big difference between being able to find instances of people doing this kind of thing, and widespread hacking of results. I think it's a given that it goes on sometimes, especially in cases where there is financial motive for doing so. But as far as it being a widespread issue? I have never seen evidence of that.
Beyond that, there's people actively working on ways to correct for publication bias in meta-analysis, to the point that it's a pretty fundamental aspect of any proper analysis.
