OHDSI Home | Forums | Wiki | Github

'Presumptive' Codes for Results


If a source EMR system has codes like ‘resumptive positive’ or ‘presumptive negative’, should they both be mapped to ‘indeterminate’? This is a tricky question to resolve as presumptive positive as has a standard SNOMED code but there is no equivalent for presumptive negative.

This is about laboratory test results, right?

During the mapping, I’d suggest to map to the most granular concepts exist. You can go with Presumptive positive and Uncertain since ‘presumptive negative’ doesn’t make a lot of sense to me.

On the phenotyping side, it depends on the granularity you’d like to achieve. Distinguishing between strong positive/weak positive/presumptive positive and positive lies somewhere in the area of specific studies in molecular biology what is not a main purpose of OMOP CDM. Most of the observational studies should be satisfied with positive/negative result, the rest if kind of junk.
But you still need to decide, whether ‘presumptive positive’ will be treated as ‘positive’ or omitted. It depends on method, analyte/pathogen, study aims, current practice in research area and investigator’s opinion.

For instance, according to FDA/CDC presumptive positive COVID-19 test result (please don’t confuse with ‘presumptive lab evidence’) is strong enough to consider the case as laboratory confirmed.

@Alexdavv, so, you are suggesting a SNOMED code as opposed to LOINC. That would explain why were were struggling; we were only looking for LOINC codes b/c the lab test is coded in LOINC. I know SNOMED is considered standard as well, but a mix of LOINC and SNOMED for an answer can be confusing. I guess we’d have to add that value in a concept set. @Christian_Reich in the COVID studies did you come across “presumptive positive”? Thanks.

Why is it confusing? LOINC by itself uses SNOMED concepts as suggested answers within the answer lists, look here.

That’s helpful. I wish LOINC added the presumptive values. From a study point of view, I’m not sure how those codes would be used. We will use SNOMED then and have to include it in our concept sets.

LOINC did, but the question is still what values were used while data generated.

Right. As for now, we have to collect all the relevant values from both LOINC and SNOMED.