I am working on a research project which receives data from an administrative database. The data comes as LOINC terms and I’m trying to match these to clinical testing.
Clinically, we would order a “screen” test, which uses immunoassay technology to identify the presence/absence of substances, either as a category (e.g. “amphetamines”) or a specific substance (e.g. “cocaine metabolite”). We want to know if the screen tests were sent for confirmatory (generally LC-MS, maybe GC-MS) testing, which is not prone to false positives/negatives.
I understand that there is a “screen” and “confirm” variable in the LOINC code, but this is not matching clinically. For example, drugs like aripiprazole or naltrexone, which I have only seen measured in mass spectroscopy (there is no corresponding immunoassay for these) only have a “screen” option, such as “ARIPiprazole:PrThr:Pt:Urine:Ord”.
Is there any way to know what the test correlates to? I know that LC-MS could also be used as “Screen” (although rarely) but generally how can I know if the test was immunoassay or somthing different? Any insights greatly appreciated.
Hi Tanya, I’m sorry for the delay in responding, had been out of town. There are two possible options of database builds covered by the screen/confirm combination vs methodless LOINCs for the DRUG/TOX class.
If the site uses two different local codes in the LIS to order up the screen and then reflex the confirm, they would use the individual screen/confirm entities within LOINC to segregate them. I believe it’s a CLIA required if something requires confirmation that it’s to be done by a different methodology. In DRUG/TOX originally, they hadn’t been identifying the different methods, but I believe over time LC-MS and GC-MS have been built into the database, as maybe IA has as well. Those individual methods refute the purpose of the screen/confirm LOINCs.
If an LIS uses only one result field for both the screen and confirm, the methodless LOINC was to be attached to that one LIS result field.
It’s entirely possible that with submissions of the mass spec and IA terms, there’s been a migration away from the identification of which one is screen and which one is confirm. If you’re dealing with one laboratory, clarification can be sought to identify which is which; but if data comes from a multitude of labs, there’s a more arduous situation. If there’s HL7 messaging, could it be asked to look further at the OBX portion to identify the methodology? Not all labs populate that portion (I’m aware).
I’ll take this as an assignment to the lab LOINC committee to make an assessment of the DRUG/TOX class and see the gap that needs to be addressed. Thank you for pointing this out. Hopefully other committee members will make more comments here for discussion.
Best regards,
Pam Banning
Hi Pam,
Thanks so much!
I am primarily working in a clinical context, so I apologize but I don’t know what LIS means?
I don’t think (if I understand it correctly) that OBX is available for me to see.
The data is coming from many different labs, so it is definitely a complex situation.
For cocaine and its metabolites, I can’t find anything that seems to represent GC-MS or LC-MS, which doesn’t make sense - I know these tests are being done.
We have lots of 19359-9 [Cocaine:ACnc:Pt:Urine:Ord:Screen] so I wonder if it is the immunoassay, but less 16448-3 [COCAINE:ACNC:PT:URINE:QN] so I wonder if it is the LC-MS, but it is hard to know?
so much appreciate any help with this,
Tanya
Hi Tanya,
LIS=Laboratory Information System (the computer system creating accessions, tracking open result fields and reporting all laboratory values)
Has the project taken the time to write the use case, the players/actors and heuristics of how the data will be mined? If the use case is strictly Drug/Tox and LOINC is the preferred terminology, it would be helpful to outline how LOINC worked with CLIA requirements in the beginning of LOINC releases, and wasn’t creating specific method terms (IA, GC, GC-MS, etc). As LOINC expanded its use case, individual methoded LOINCs were submitted for (research, non-medical uses).
I would NOT draw an inference that methodless 16448-3 is LC-MS. The original intent had been that methodless LOINC is a combination of screen and confirm when the LIS only has one result field per drug. Any methodless LOINC is not a specific method. The HL7 v2.5 OBX-17 segment is optional to record method.
The heuristics you create will document for others the possible shortcomings of incoming data, inaccessibility to HL7 optional OBX-17, inability to segregate and count IA to LC-MS to GC etc.
Where shortcomings are found, it’s helpful to share the heuristic document and get the brainstormers in to create a process improvement reiteration of the project. Or at the very least, introduce the shortcomings to the incoming data streams and ask for improvement.
I may be taking too high of a path to possible solutions, but it never hurts to ask.
Have a great weekend,
Pam
Concur with Pam, that there is not necessarily a direct correlation with the test method (IA, LCMS, GCMS) and whether it is screenign or confirmatory. As noted, some drug testing does not have a commercial immunoassay and may only be done on LCMS or GCMS. One reason is it is not commonly performed and thus a commercial vendor has not developed an immunoassy that has 510k FDA approval to be purchased. Larger laboratories may “specialize” in providing testing for some of these drugs too. Often they will be a lab developed test and thus on GCMS, LCMS. If they have a high volume of some drug tests, it may make sense financially to develop a LDT in house even if there is a commercial test available too.
That said, drug testing is designed for a variety of purposes: screening, confirmatory, emergency room overdoses, drug rehab program monitoring, newborn exposures, treatment monitoring, metabolite detection, autopsies, environmental/occupational exposures, monitoring of transportation/nuclear industry employees, etc.
