If this is a one-time thing you might find this to be one usable
alternative - assuming you've already scanned the forms as images and they
are in word docs:
In one file at a time, Select & Cut the image then go to the Header/Footer &
paste the image there. Position it as precisely as you can then go back to
the text layer of the doc. (I'm suggesting the Header/Footer because if you
simply format the form image as Behind Text you run the risk of moving it
accidentally as you continue to do the rest of the work. Also, Text Boxes
because trying to get regular typed text to go into the right spots can be a
severe challenge.) Don't be concerned that the form image appears to be
"faded" once you leave the H/F layer - it will print with full intensity.
In each location where you need to enter information, draw a text box
formatted with No Fill & No Line, Text Wrapping set to In Front of Text,
whatever font specs you need & type the info into each box as you go.
Graham is quite right about the OCR (Optical Character Recognition)
approach, but to get editable text you lose all the graphic elements (such
as lines or logos & seals). As he wrote, a veritable nightmare to deal with.