Comparison Studies: Acceptable differences between Coaguchek XS Pro and Stago

So one of our clinics use an CoaguCheck XS they had used a difference of INR of 0.5. I started looking into this recently and found that there appears to be a bias in the 3.0 to 4.0 range and the samples don't match that closely. When looking at acceptable limits of proficiency testing it said INR was +/- 2 SD. So I am still trying to find a happy middle ground. 

We have varying parameters depending on the INR range since there is fluctuation bias.
+/- 0.3 for INR less than 2.5
+/- 1.0 for INR equal or greater than 2.5
+/- 2.0 for INR equal or greater than 4.5
I called CAP about this issue. She told me not to stress about this...it's comparing apples to oranges. She said it's fine to review historical data to determine acceptable limits. As an example, the Stago results are x% lower than Coaguchek, and to adjust our acceptable parameters to reflect the x% difference.
We have new Coaguchek analyzers, and the comparison was redone, so I'm hoping they may show more consistency. 
I was curious what others found with their comparisons, and if we had similar parameters
Thank you for your response!

@Liz Reed, how do you adjust for the x% of acceptable range? Do you have a standby in SOP that it is acceptable to have the x% of variation between Stago and Coagucheck? If yes, then what resource are you using to back up the statement? Is there a ACAP document floating somewhere that indicates that x% is acceptable? As you noted the x varies with increasing INR

We are about to start our comparison as well for the Coagucheck against Stago. I am curious how to support the variations as we tried to validate about 7 years ago.  

Diana, I'm basing the x% adjustments on what CAP told me. For CAP/CMS compliance, she said all we are required to do is establish a range between the instruments via historical data and use that as our parameter since they aren't perfect numbers especially if the patient populations are different (we use different instruments at different sites - patient population is not the same for each site). She said the purpose is to be able to explain to a physician why a lab PT/INR may not correlate well with a POC instrument. 
On our comparison report, we will put the acceptable limit ranges determined by historical data. 
I don't have a backup statement in our QAP. I've been all over the internet trying to find information regarding the differences but with little luck. That's why I consulted CAP for guidance. I also should have qualified that we measure PT seconds by % difference and INR by the ranges listed in the previous comment. The INR ranges came from our medical director at our Anti coagulation Management Services. The literature he used came from Roche's Coaguchek website indicating the higher the INR, the greater the deviation. He also noted this trend in our own patient studies.
As far as supporting the variations, if you find something in literature, please add it to this thread. Honestly, I'm stumped.

We use 20% difference.  Roche Coaguchek say they allow up to 30% difference, but my medical director wanted it tighter than that and we set it at 20%.

@Liz Reed, thank you very much for the detailed explanation. If I will come across supporting literature, I will share it in the thread of this post. 

YW... and thanks for sharing any future literature.

I think I'll stick with the Coaguchek Plus after reading this. I really don't want to try to compare bedside INRs to our ACL TOP :(

Ken, it's our PT seconds that is the issue. Since we have been reporting it out, we are required to perform comparisons...which is basically comparing apples to oranges. Our clinicians don't use the PT sec, only the INR so we're in the process of removing the PT sec from reporting out. We've had no trouble with that comparison.