Our Website

Check out our website for even more up to date information!!

Thursday, October 25, 2012

How accurate are INR tests between different systems?

It has always seemed to me to be very hard to find out just how accurate INR tests from drawn blood taken at pathology labs are - for example, if they indicate your INR is, say 2.5, is it really between 2.3 and 2.7, or 2.45 and 2.55 - that is, in technical terms  what is the standard deviation of the test  or just how accurate are they.

Well I have found a study comparing INR as determined on the CoaguChek S (Roche Diagnostics, Indianapolis, IN), CoaguChek XS (Roche Diagnostics), and i-STAT 1 (i-STAT, East Windsor, NJ) point-of-care (POC) analyzers compared with venous plasma INRs determined by a reference laboratory method.

The abstract of the paper said

" Overall agreement between POC (Point Of Care) and laboratory plasma INR was very good, with median bias between capillary whole blood and laboratory plasma INRs varying from 0.0 to –0.2 INR units on all devices. More than 90% of results on the CoaguChek XS and i-STAT 1 and 88% of CoaguChek S results were within 0.4 INR units of the reference laboratory method. The CoaguChek XS and i-STAT 1 demonstrated greater accuracy than the CoaguChek S as measured by the number of results that differed by more than 0.5 INR units from the reference method. Median bias between CoaguChek S capillary whole blood and laboratory plasma INRs changed over time, demonstrating the need for ongoing quality assurance measures for POC INR programs."

Sounds heavy stuff.

But at least there is some indication that there can be differences between the various systems  sometimes up to 0.5 INR units or more. Further there is some indication that over a period of months, the relationship between the various systems changed.

But the problem is that I still have not been able to find out just how reliable or consistent the INR method relying on drawn blood by a pathology lab is. All this paper does is to compare various systems for measuring INR without indicating the reliability of each method in returning consistent results from a single sample. The article is well worth reading, and you can find it here

No comments: