The basic problem is that scanning, ocr & conversion do not give you a clean Word document. They give you a formatting mess that may look fine. When you try to edit it, things can get confusing. This is true of all conversion and ocr processes of which I am aware. There have been great advances in this technology but I would be surprised if it ever becomes perfect.

Generally, when I want to use a scanned document (or text from the web) to produce an editable document, I do the ocr and copy into a new Word document as plain text. I do the formatting there. Note, also, that the best ocr is still not 100% accurate. There can be errors that are not apparent to the eye like the substitution of the number 1 for the letter l or vice versa. The blurrier the original, the more the errors.