Register Login

Multi-site Reference Intervals

While attending the September 12–15, 2017 Mayo Medical Laboratory conference, Bleeding and Thrombosing Diseases George spoke with Susan Hollister and Jennifer Rider from Sanford Health in Fargo.

Susan and Jen are developing reference intervals for their coagulation laboratory services throughout Sanford’s system and are looking for a reasonable approach that combines satellite resources with the central laboratory service. George referred their question to Heather DeVries, Indiana University Health Center, who manages a large network. Heather’s response was…

Normals are a big issue for the satellites. I would first refer to CLSI document H57-A, a protocol for evaluating/validating/implementing coagulometers. In a section on reference intervals, the document states that “testing conditions should replicate how the laboratory functions, eg, many hospital laboratories use fresh plasma for all testing and evaluate performance on fresh plasma; whereas some laboratories receive virtually no fresh samples and therefore validate everything using frozen/thawed samples.”

Also, reference intervals should be created from the local population, but I understand how that can be very difficult for the satellites, especially when they need 20 for a geomean. We send out only a few frozen normals to our satellites, and ask that they do whatever they can to get fresh. Our lab gives out meal tickets!  Also, I remind the labs that they can contact nursing managers, pharmacy, etc., for help   It doesn’t have to be lab personnel only that is part of the reference range. I may also, as a third choice, purchase plasma sets from trusted  vendors like Precision BioLogic Inc. or George King BioMedical Hope that helps!

George reviewed Susan, Jen and Heather’s comments with Andrew Goodwin, MD, University of Vermont Medical Center who confirmed Heather’s suggestions.

George adds that most systems prefer to develop a single RI throughout the system for each assay for portability. Patients often move among satellites or are transferred to the central service, and variant RIs create clinical confusion. This requires a mathmatical approach to combining RIs from each facility. This approach is feasable when the facilities all use similar methods.

For discussion’s sake, is all this effort necessary, or could there be a simpler approach, for instance, the judicious application of published RIs? Dr. Ken Mann, who spoke at the Mayo Conference, makes the point, in hemostasis, there is so much inter-patient variability, the concept of an RI may not actually work particularly well. I’d like to “hear” from some of our participants on the topic of RIs.


Comments (2)
Apr 4, 2019 9:40am

I was one of the authors of
I was one of the authors of CLSI Document H57-A. I also conducted a little experiment in my own facility where I compared like specimens both fresh and then frozen at -70C and thawed one time at 37C. I ran PT/INR, PTT, and fibrinogen levels on our instrumentation. My findings showed no clinical or statistical difference if the specimens were processed properly. Specimen handling is where differences seem to occur among sites. You might also check McGlasson DL, Necessity for proper calibration of reagents and instruments in coagulation testing networks. Lab Medicine. 2004;35:461. The following might be of some use: McGlasson D, A comparison of INRs after local calibration of thromboplastin international sensitivity indices. Clin Lab Sci. 2002;152:91–5. We used 12 different reagent/instrumentation combinations.

Mar 25, 2019 1:13pm

Hello George!

Hello George!
I would like to revisit this post, as our system would like to try and use a combined INR reference AND reportable range. With calculating instrument- and site-specific geomeans for so long, I am struggling with this. Any advice?

Many thanks,

Leave a Reply