Thank you so much for a wonderful presentation!!!
Questions:
Hung: For LIDR we are looking at how to represent the harmonization of a lab test - we will need to get help to properly cover the temporal variation?
having indicator if an assay has been harmonized is good
even a test that has stable bias the data can still be used - but you hve to normalize the data afterwards, if those data are known about the tests
Walter: Scope of the problem - does this cover all lab tests, or do we know what proportion of the lab tests have these issues?
Studies are done mostly in the chemical lab domain, but there are limited resources to testing across the globe, so we are focusing on the clinically important
could set up the clinical study in a way to minimize some of the bias
the issues normally only show up when there are clinical indications of issues are reported
Andrea: Is there a way to group some of the measurants into categories?
harmonization complete
no harmonization neeed
unknow status
Other considerations:
Would not recommend the regular profiecincy studies for comparability across institutions
Consider that clinical guidelines change over time, which might require adjustments to the studies, as the expected concentration targets have moved, so check that the tests still perform well enough at these lower target values
Need to ensure we address systematic errors
Walter: what is the relative role of the maufacutrers? Can the lack of stnadardixation be addressed by the manufacturers, or is it on the lab side, or does it vary analyte by analyte
calibration bias must be addressed on the manufacturer side - that may even require a NEW FDA approval
Any issues in the genomics side (Epstein Barr testing article: Rychert J, Danziger-Isakov L, Yen-Lieberman B, Storch G, Buller R, Sweet SC, Mehta AK, Cheeseman JA, Heeger P, Rosenberg ES, Fishman JA. Multicenter comparison of laboratory performance in cytomegalovirus and Epstein-Barr virus viral load testing using international standards. Clin Transplant. 2014 Dec;28(12):1416-23.)?
Next Gen Sequencing are high on the radar, since range is wide and use steadily increading - but so far no programmatic approach for this yet as it requires NEW techinology to address those, so focusing on traditional chemical analytes
Hung: LIDR collects attributes covered by different SDOs - would the curation of a harmonization indicator fal into a different orgnaization - hard to figure out how to address the temporal
the manufacturers and labs participating in the program are on website nd could add coding to that website on how to communicate the temporal
Andrea: Would some of the calculated resutls be part of that - where calculations could address some of the biases (lso in the chat)?
tests not done by commercial manufacturer are out of scope
For challenge tsts and other clinical context assume using order LOINC
could we add modifiers to tests to flag certain results when used for a differnt population with adjusted reference range
there are curently 3 different formulas for LDL - would be good to get a handle on that
Walter: Standardizarion around INR - is this part of this space?
Normalization and adjustment factors
have gotten a lot of questions around usefulness of data of older studies like the Framingham study - assess for each situation what adjustment might be usable in that situation
Andrea Pitkus, PhD, MLS(ASCP)CE, FAMIA. UW-Madison 12:31 PM
Thank you, Dr. Vesper, for an excellent presentation!
Murray, Bill to Everyone 12:32 PM
great presentation!
Andrea Pitkus, PhD, MLS(ASCP)CE, FAMIA. UW-Madison to Everyone 12:35 PM
https://www.harmonization.net/measurands/
Xavier Gansel to Everyone 12:38 PM
I have to drop off, I appologize for that
Andrea Pitkus, PhD, MLS(ASCP)CE, FAMIA. UW-Madison 12:38 PM
How are you using the term "test?" Panels or individual analytes, etc?
Scott Campbell 12:48 PM
Hi Riki, where will the recording be posted?
I was obviously very late…blame the shingles vax
Andrea Pitkus, PhD, MLS(ASCP)CE, FAMIA. UW-Madison 12:50 PM
As a follow up, Dr. Luu, are you also thinking timed, calculated results and values would be part of that or separate as manufacturers are unaware of how these are set up by labs in their LIS?