Strange idea, but looks pretty disturbing. Please see my comment to the question.
Let's see if it is possible to investigate.
The *.docx document format is Open Office XML, it is open and standardized under ECMA-376 and ISO/IEC 29500, please see
http://en.wikipedia.org/wiki/Docx[
^].
That said, you can obtain the standard and see what can it contain in principle. It contains some meta-data, and, if nothing is encrypted (I hope), you can rename the *.docx to *.zip, unzip the files and perform the search.
The situation with *.doc files is much more difficult (but you don't have to use them, you can save everything as *.docx). There are several different versions of the format; they are all proprietary. Nevertheless, Apache OpenOffice can read them all, so you can download its source code and see how Word documents are parsed:
http://en.wikipedia.org/wiki/Open_Office[
^],
http://incubator.apache.org/openofficeorg/source.html[
^].
You will need a Subversion client to download the source code, which is not a problem at all:
http://en.wikipedia.org/wiki/Subversion_%28software%29[
^],
http://subversion.apache.org/[
^].
The only problem is: OpenOffice may or may not parse or interpret all of the data. Anyway, you can figure this out of the source code.
Care to try your investigation? In you do, please report your results, write an article.
—SA