Some points to consider:
1. In most cases PDFs such as the one you're trying to recreate as a Word document can be exported from the PDF as a Word document, using products such as Adobe Acrobat Pro (v.8 is available as a free download from
http://www.techspot.com/downloads/46...at-8-free.html - note the serial# mentioned there), without the need for an OCR process.
2. From a cursory check, it appears some or all of the documents are available in such a PDF format at:
http://www.escambiaclerk.com/clerk/coc_evictions.aspx
3. If you contact the people who produced your PDFs, they may be willing to provide a copy in a Word-compatible format, which would obviate the need for you to work with PDF conversions. The PDFs at the above site, for example, were produced using WordPerfect and Word can open those files.
4. Your OCR software has apparently tried to replicate some of the document structure via a multi-column page layout, whereas a multi-column table may have been more appropriate in some cases.
5. It is not clear that you need formfields to make the document usable as a form. The long-address issue, for example, could be handled via the use of a table cell with a fixed height and width, sufficient to accommodate the longest addresses. With such a cell, the page layout won't be dependent on the length of the address. In any event, you may find content controls more flexible than formfields.
6. You also mention moving the final product back to Acrobat. Neither formfields nor content controls will survive the conversion - PDF forms use an entirely different kind of field, unique to that format.