Hung Luu

Children’s

Riki Merrick

Vernetzt, APHL

Andrea Pitkus

UW

Pam Banning

3M - Solventum

Xavier Gansel

Biomerieux

Amy McCormick

Epic

Dan Rutz

Epic

Rob Rae

CAP

Rob Hausam

Hausam Consulting

Sandy Jones

CDC

Stan Huff

Graphite

Ed Heierman

Abbott / IICC

Andrew Quinn

Laurent Lardin

Biomerieux

Anthony Killeen

UMN

Craig Collom

Marti Velezis

 Sonrisa / FDA

Walter Sujansky

FDA

Susan Downer

JMC

Ralf Herzog

Roche

Cornelia Felder

Roche

Daniel Golson

JMC

Andrea Prada

JMC

Maria Sagat

 CAP

Raja Cholan

FDA

Russ Ott

FDA

Akila Namasivayam

FDA

Desiree Mustaquim

CDC

Reviewing minutes from the last call - Action Item Follow up

• Pull out the definitions from LIVD for test kit and equipment and put here: SHIELD Glossary

  • Review operational definition of “equivalence”

• Outreach to Dr. DeBaca

  • Hung will check for more info and maybe see, if she could participate in SHIELD

• Waiting for response from Ed

• Keep updating the googlesheet: Potassium LIVD data

• Shared Reportable Conditions LIVD google sheet for crowdsourcing with CSTE - added sign up sheet for PHAs to specific conditions and see how that develops

• Have LIVD File Repository Requirements

  Have draft budget (for setting up a web-based database and yearly maintenance): https://aphlinformatics.atlassian.net/wiki/download/attachments/915407143/LIDR - Budget.xlsx?api=v2

• Sign up for sections to provide draft content on the LIDR White Paper

• Review LIDR Use Cases and sign up to lead one of these

• Define next steps to migrate existing LIVD files into FHIR LIVD catalog format and find a FHIR server to host

Call Schedule

send OOO via chat or email

LIDR Elements Discussion

Review the PT and Haemtology example Hung created, with comments from Andrea

1. Most of the automated Hem analyzers analyze both tubes of whole blood obtained by venipuncture and microtainers from capillary collection.  May need to add those details. 

2. A number of the order LOINCs are missing (e.g. H&H, CBC W/O diff).  In some cases, there are 20+.  Give the large number of additional rows to support, I lean towards excluding orders/order LOINCs.

3. For a number of the reagents (test kit), I noticed some have different sizes and thus different identifiers for each.  Also some appear to be missing where multiple reagents are used together to produce a result value.  I don't have a user guide so not sure if differences across analyzer models too.

4. Did literature search for comparisons.  Added columns for the citations, FDA documentation and also which results might be considered "equivalent."  

5. Added Specimen Collection Procedure and moved Specimen Source terms/codes to that column as that is what they are.  It's great to see the codes as that is really helpful. (and glad not the combinatorial explosion we see with Specimen Cross Map Table!)

6. For the iStat Chem 8 Cartridge, I disagree with the LOINC map as the indicated documentation indicates this is a calculated Hgb.  There is a different LOINC available for the calculated method.  Also why I gave it, it's own group.  (There are 2 different cartridges for iStat CO2 where one is calculated and one is not, and thus 2 different LOINCs are warranted for those too.)

7. I added colorization to the WBC and Retic rows to visually distinguish that they are different from the Hgbs.

8. The Hemocue literature indicates Hgb values are plus minus 7% bias with the automated heme analyzers.  It appears to be within an acceptable range for CAP, CLSI.  That begs the question as to how much is an acceptable difference in result values by different instruments, methods, reagents, etc.  Given different reference ranges for each result, may need to decide on an individual basis as 7% may be too much for something with a narrow range, but acceptable for something with a wider range.

9. Where different specimens are allowed, the question we consider with mapping is what is built as the result in the LIS?  Is it a single Hgb built even though capillary or venous, or arterial or venous specimens may be analyzed and reported?  If so, the whole blood Hgb is usually the best map.  If separate results are built similar to Arterial and Venous Blood Gas panels, then a LOINC system specific to the specimen may be warranted.  I want to check with Pam on which way she leans in these cases.  I note I tend to lean with the whole blood Hgb for the POCT testing, and use the Arterial Hgb, Venous Hgb to reflect syringe collections for more blood gas panels to distinguish the two types of tests.  When a generic Blood gas panel is built and used for the different types of specimens mapping to the arterial hgb or venous hgb isn't possible and the pendulum swings towards the whole blood hgb. 

10. I'm ok with the specimen of whole blood with edetic acid for now.  Riki is aware there has been work started on aspects such as container info (type like glass/plastic, size, vol, gel/no gel, etc.) as well as additives (EDTA, %, dry, liquid, preservative, etc).  It could arise in discussion. We may want to park it for now.  It would be another layer of combinatorial explosion.  As we know for pathology, and certain tests, it is vital info to know.  HL7 specimen call has fields to collect I think most of this, as it comes up from time to time, but I don't believe complete assessment has been done (correct me if I'm wrong Riki) of all the different aspects and it SCT has them all represented  Is this something you are both expecting in LIDR or not?.Or on a case by case basis?  

Discussion:

Pro/Con of using the methodless LOINC: having a single code may be beneficial to make it easiest for the lab to pick the LOINC- as long as they can send the specimen and the device identifiers.

To get the WBC you need multiple reagents, some of which are preparations, but we should focus on the reagent kit that creates the numeric values, rather than all the step

Question is how to decide which reagent is being reported - we need clear definitions: the one that creates the result

Ratio is a different situation, as that is a calculation

When is the reagent important?

When a given method first comes out, differences in results are often observed - but over time often these disappear, so focus not so much on the reagents

For hematology the reagents are harmonized

Map uses this information to map thier set up, so we need to esnure we include all possible combinations of the reagents that are important - ensure we limit to the

CLIA describes the test system - it is based on the package insert, so if the package insert includes mulitple choices all should be represented

Most labs set this up using whole blood in closed mode, but for some samples (hard stick, pediatric patient) is run on fingerstick in open mode - in the labs this is often still set up as a single test in the LIS - so would need just one LOINC - whole blood LOINC needs to be used, and then include the specimen type

Creating grouping

3 groups

Row 7 Abbott iStat: includes a calculation and best practice for mapping would be to use the calculation LOINC instead.

Hung: If we ensure we include the additional elements then we could use a more broad LOINC.

Walter: If the test actually is a calculation should really highlight the difference to suppport secondary data use.

Pam: Often looks for settign up catalogs for iStat - often set up different from others in the LIS, sometimes using arterial blood - to help surgeon

all CBCs are also done on venous blood - should include that

may need to be able to set up LOINC by analyzer - LIS should be able to keep track of that

Have not seen collection procedures used for specimen source - we can include that as another column

Equipment and reagent exists, but not neccessarily the device ID, so we should see, if we can get the vendors to submit to GUUID for them

sometimes there are sizes/volumns of the reagent cubes have different UDIs for the same reagent - should we then have 3 different rows? Hoping that we can get the prodcution IDs in the instnace data, but until we can get there we should probably include these different sizes (HemoCue for fingerstick is combining prep insided the collection device and it is often not interfaced) - so for this then we would use only the device identifier for the instrument

Stan - Postcoordination and using generic LOINC: If you include the method and instrument in the data you can then use that broader LOINC for queries and access, if the other elements are incuded, then the researcher can review them and excude or include as appropriate

these details are probably never clinically relevant - focus only on where we know there are differences

From equivalence standpoint: Rows 3, 4, 5 are one group, and then row 6 is separate

hemoglobin is listed as harmonized - we can check with Hubert on feedback

Next call take up these questions:

• referenece range differences - what is acceptable

• discuss approach about broad LOINC vs more specific LOINC

• Keep iStat separate - since POCT device?

Review LIDR White Paper

 not discussed

ACTION ITEMS

Please see the action items at top of this page - Next deliverable is White paper draft by end of this month

And we need to prioritize the use cases, so that we can finalize the requirements for the first phase of LIDR, which need to be included in the White paper

Next call

Monday 3/4/2024 9 - 10 AM ET

Adjourned

10:01 AM ET