Any possibility of having your PAs and/or pathologists switch to using voice recognition software and eliminating the need for transcriptionists altogether? I wasn't here for the changeover at my current group (which I heard was painful, as most major changes are), but our current system is incredibly efficient. PAs all use roughly synoptic format gross description templates for common specimen types, so all they have to do is fill in sizes/measurements and brief descriptions - also makes the grosses really quick and easy to read. We all use Dragon to generate our reports as well with varying degrees of automation and voice command macros (can autogenerate the specimen designations/headers for most derm, GI, prostate biopsies, etc; we have lots of macros for frequently used diagnoses and/or canned comments, synoptics, etc. as well as to do things like document intradepartmental peer reviews, open the EMR and look up the current patient, call the ordering physician, put various notes into the case, etc.).
I suppose it might not be worth it if you already have an arrangement with those hospital system clients to provide you with transcription services at no cost to you. Can also be a bit difficult if you have pathologist(s) in your group who struggle with reviewing their own reports for voice recognition errors/typos and previously relied heavily on transcription staff to do that for them.