I have. According to [1], "1 in 4 cancer research papers contains faked data". As the article argues, the standards are perhaps unreasonably strict, but even by more favorable criteria, 1/8 of the papers contained faked data. Interestingly, [2] using the same approach, found fraud in 12.4% of papers in the International Journal of Oncology. More broadly, [2] found fraud in about 4% of the papers studied (782 of 20,621). I'd say that's pretty widespread, but you further have to consider that these papers focused narrowly on a very specific type of fraud that is easy to detect (image duplication), so we would expect the true number of fraudulent papers to be much higher.

[1] https://arstechnica.com/science/2015/06/study-claims-1-in-4-...

[2] https://www.biorxiv.org/content/biorxiv/early/2016/04/20/049...