It only applies to FAT32 formatted hard drives. Regular E-mu EOS formatted drives can be in the chain just fine, because the computer won't see a validly formatted hard drive (and, hopefully, will just ignore it... although it's always hard to tell with Windows operating systems because Microsoft insists that it knows better than the users what the users really want, and frequently tries to be "helpful" at the most inconvenient times)
Oh, I should probably stop ranting... (and excuse the lengthy posts, but I'm a writer and tend to get carried away with my words)
The main worry is having the computer, as a master device, and the sampler, as another master device, trying to address the hard drive at the same time. SCSI doesn't normally allow more than one master device on a chain at the same time, and the conflict is what might cause the data corruption that the manual refers to (as the hard drive gets confused about who's in charge and tries to kill itself... hard drives are notoriously emotional things, and you really need to be gentle with them or they get scared). EOS works around this because essentially the EOS formatted hard drive becomes an un-readable device as far as the computer is concerned and it just ignores it (the sampler puts on a hat that says "I'm not a master device", the computer is happy in its ignorence, and the SCSI hardware in the external drive box just routes SCSI transmissions from the computer through to the sampler as if the drive wasn't even there, while the sampler can still talk to the external drive box just fine).
The down side is that the EOS formatted drives can only address a maximum of 18GB of drive space.
Do you already have an internal drive in the e5000? If not, you might try getting an IDE drive for the inside, since the IDE bus isn't shared at all (the computer couldn't see the drive even if it wanted to, so no conflict). Then the internal drive could be formatted for FAT32 and be much larger than 18GB (I've got a FAT32 formatted 40Gb IDE drive in my e6400-Ultra). Unfortunately, the internal SCSI bus is on the same chain as the external SCSI connection, so you'd run into the same problems as with an external SCSI drive.
There is a chance that everything would be just fine, as long as nothing on the computer tried to access the FAT32 external drive while the sampler is accessing it (so turning off auto-update things and such). It's when the drive gets two access requests from different masters at the same time that things go all apocalyptic. **PLEASE** Do not take my word for this, as my grasp of the intricacies of the SCSI protocol are sketchy at best. This is why I use words like "a chance", any risk is yours to take, you have been warned, etc...