OHDSI Home | Forums | Wiki | Github

Glucose [Mass/volume] concept - mapping source values for cohort definition

Researchers
, ,
#1

Hi all, we want to create a cohort definition involving glucose measurement using threshold values - which are different for fasting glucose measurement and random glucose measurement (in serum/plasma). The standard concept for measurement of “Glucose [Mass/volume] in Serum or Plasma” (concept ID = 3004501, LOINC concept code = 2345-7) in our EHR dataset has multiple measurement_source_value(s) with the same measurement_source_concept_id (3004501) - as below. There are about 65 distinct measurement_source_value, I have listed the ones with high record counts in our dataset.

BASIC METABOLIC PANELGLUCOSE
GLUCOSE
COMP METABOLIC PANELGLUCOSE
GLUCOSEGLUCOSE
GLUCOSE (OO)GLUCOSE
RENAL FUNCTION PANEL (OO)GLUCOSE
CMP & GFR(QUEST ONLY)GLUCOSE
COMP METABOLIC PANEL W/EGFRGLUCOSE
BASIC METABOLIC PANEL W/EGFRGLUCOSE
RENAL TRANSPLANT POST-OP LONG LAB SETGLUCOSE
FASTING GLUCOSEGLUCOSE
LIVER TRANSPLANT POST-OP LONG LAB SETGLUCOSE
CMP W/EGFRGLUCOSE
BMP & GFR(QUEST ONLY)GLUCOSE
RANDOM GLUCOSE (OO)GLUCOSE
ADD ON BMP W/EGFRGLUCOSE
ADD ON CMP W/EGFRGLUCOSE
RENAL TRANSPLANT POST-OP SHORT LAB SETGLUCOSE
COMMUNITY HEALTH SCREEN W/LDLGLUCOSE
COMPREHENSIVE METABOLIC PANEL W/EGFRGLUCOSE

As seen above, the measurement_concept_id (3004501) includes many different source values. ATLAS allows using measurement_source_concept_id to create the cohort, however in this case it is the same for all values. To use the correct thresholds we need to be able to differentiate between them and remove ambiguities such as ‘GLUCOSE’ with no other specification. Wondering if someone has looked into this and used these concepts previously, or if there is some way to alter mappings and use the measurement_source_value. Thank you!

#2

@sanyabt:

These are all Glucose measurements it seems. The fact that they are given in panels, with other measurements or to certain patients (renal transplant) is irrelevant. OMOP is patient-centric, and those distinctions are relevant for running the hospital, not the diagnostic.

Only exception is fasting glucose, which has its own LOINC.

What’s your use case? Why do you need to differentiate?

#3

@Christian_Reich thank you for the response! We are attempting to create a cohort definition to identify patients with diabetes using both the diagnostic codes and glucose measurement values (in a specified time period). I could modify the fasting glucose concept ID above, but we are also unsure if the other measurements of glucose (BASIC METABOLIC PANELGLUCOSE, GLUCOSE, COMP METABOLIC PANELGLUCOSE, GLUCOSEGLUCOSE) correspond to fasting or random glucose measurement as that is not specified in the codes. Differentiating is required as random and fasting would have different thresholds for identifying diabetes.

#4

Totally understand. But it’s the job of the ETLer to figure out how to interpret these local codes. We have no way of knowing what “COMP METABOLIC PANEL GLUCOSE” means. You got to ask your local people who know this.

t