Lab test mapping to LOINC

Implementers
1

Hi all,
LOINC has several codes for measurement of the same substance regarding the circumstances. Few examples for blood glucose;

2339-0 Glucose [Mass/volume] in Blood
41652-9 Glucose [Mass/volume] in Venous blood
1557-8 Fasting glucose [Mass/volume] in Venous blood
2340-8 Glucose [Mass/volume] in Blood by Automated test strip

For a test named glucose level in the source data; it’s known that, it’s almost always measured from venous blood, unless it’s a strip test, and it’s mostly fasting glucose level. But, for sure, this is an assumption, since it’s not explicitly referred by the test name. So, in that case, whatever the code we select, it will be wrong. Besides this data may already include different methodologies like automated tests since data sources are diverse.

Is there a general policy regarding this kind of encoding problems, like “select the most generic one”?
Thank you for your time,
Ergin Soysal
@UTHealth

2

@ergin:

Yeah. This is a problem we know. LOINC is very detailed, usually much more detailed than the source data allow you to determine.

Here is what you can do: You can use the hierarchy that LOINC provides. If you go up you will find LP42107-0 ‘Glucose | Bld-Ser-Plas’. The children of this are all the various measurements of Glucose in the blood or its derivatives. However, LP42107-0 is standard_concept = ‘C’, which means it is a classification and should not appear in the CDM data table. So, I would pick 2339-0 ‘Glucose [Mass/volume] in Blood’ and rely on the analyst to use the hierarchy in the CONCEPT_ANCESTOR table to find it.

Does that help?

1 Like
3

Thank you.
-e.