Multiple slicers and information accuracy

Hi,

We have a report that we're trying to put together that originates from 3 different sources.

1. Budget file, includes expected patient volume by surgeon, and includes surgeon specialty
2. Patient file, includes patient level data, including a patient ID, expected revenue, etc.
3. OR actuals file, also includes the patient IDs, surgeon, specialty, setting (Inpatient or Outpatient), etc.

At the moment we're running into an issue with how the data comes together to roll up from surgeon -> specialty -> expected revenue impact, depending on the various slicers selected. 

I'll try to demonstrate the issue using only the Inpatient (IP) and Outpatient (OP) slicer:

First is "all'in" view:

Selecting IP only:

Selecting OP only:

You see that the issue is that the "Expected Rev. Impact" card doesn't make sense if you sum the IP and OP results together. 
389k + -338k <> -308k

The card is simply presenting the result of a measure that's looking at the "Variance to Budget" count and multiplying that against the "Exp. Rev/Case". And that Exp. Rev/Case is just the average of all expected payments within the selected slicer settings. What that card needs to show, is to show a sum of the average expected payments by surgeon...and roll up from there. And it needs to be able to do that based on whatever the slicer input combination might be.

Sorry if this isn't super clear, but hopefully it's a good starting point for seeking some help!

Thanks!