Jim's right, it's most likely not the patch but the destination.
I'd get the other end of the patch: either the channel on an extender /or 651 Raised, and with a multimeter check continuity from the end of a bantam lead, through the patchpoint, and on to the destination.
Chances are, as Jim pointed out, the other end is most likely a 5534 de-balancing (or balancing in the case of a sending patchpoint)
Swap out the 5534, and you'd be very unlucky to not solve the issue.