You may receive an OCRed document for translation that, when pulled into a CAT tool, contains a lot of inline tags that only refer to font size or font format (bold, italics, etc). Those can be considered unnecessary and more of a hindrance for the translator, making their work very tedious, difficult and possibly causing delays. Excessive code can also lead to inaccurate translation memory leverage, which is never a good idea when trying to optimise projects.
If you know that those tags originated from an extra step somewhere along the way (such as converting a PDF to Word), there are ways to get rid of them. One example is Codezapper; a set of macros for Word that gets rid of useless code in a flash. And there are others.
If your OCRed source document contains an overwhelming amount of tags, try experimenting with these.