Conversion from any non-Word format results in formatting and editing anomalies. Andrew has suggested a very good way to get rid of the different page orientations or sizes.

If there are strange frames or textboxes or random changes in fonts I would suggest also selecting everything and pasting into a new document as plain text and then using styles to format that text to match the original documents (not the OCR version, but the paper documents). Images can be copied and pasted from the OCR version.

OCR today is much better than it was twenty or even ten years ago but it is imperfect. Many erroneous interpretations can be caught by spell check but human proofreading is still required.