There are a few things to be addressed here:
Firstly, there are two types of non-exclusive output. There is DirectSound and WASAPI.
DirectSound will accept all sample rates that you send it, and Windows will resample it to match the output rate set for your device in the Control Panel.
WASAPI will reject anything but the output rate set for your device in the Control Panel. The player has to do the resampling.
If you use WASAPI Exclusive mode, it allows the player to switch the sample rate being sent to the DAC.
In your case, the DAC only supports 44.1/48kHz, so changing to exclusive mode would only allow it to switch between those two sample rates.
If you're mostly playing CD-quality music, I might suggest setting the DAC 44.1kHz rather than 48 in the Control Panel.
Since you aren't just sending it audio from Media Center, you probably want to leave it in a non-exclusive mode. I'd suggest WASAPI, with Media Center handling the resampling.
Now as I understand it, your setup is that you are using three AVRs to send audio to six different rooms.
While the AVRs will accept a digital input (either 24/96 or 24/192) they cannot route this audio to a second zone. They can only send an analog input to the second zone.
So what you're doing is sending the audio to an external DAC, and either splitting or daisy-chaining the analog output from this to all three AVRs, which are in turn sending that to two zones each.
This means that one input gets sent to all six rooms.
You could replace the DAC with a newer model that supports high resolutions (while it accepts a 24-bit input, the DAC itself is 20-bit) but honestly I don't see much reason to if you are happy with the sound quality.
Now something that I don't know is whether the AVRs can only split a signal to two zones, or if they can route different inputs to each - as long as zone 2's is an analog input.
If they can do this, you have another option.
If you replaced the DAC with a multichannel one, you would be able to set up six different zones in Media Center, which would give you independent audio control for each room, rather than splitting the audio from a single input across all of them.
At worst, you would have independent control over each AVR (if they take one input and split it) rather than six. (if they can send different inputs to each zone)
Of course, that's only if it's a feature that you want to use. If you are happy sending one input to all six rooms, I'd leave things as they are.