OHDSI Home | Forums | Wiki | Github

LOINC: suggestion to change concept names for concepts indicating history


(Polina Talapova) #1

Inasmuch as domains of LOINC concepts are going to be changed according to the new logic of the domain distribution, some names of LOINC concepts indicating historical facts and requiring the ‘Observation’ domain also should be modified. All of them contain ‘Hx’ as a value of a ‘property’ filed in the source (e.g. 66678-4 Diabetes [PhenX]), but do not have words emphasizing connection with a history inside the names.

We suggest to modify their concept names in the next way:

LOINC old_concept_name => LOINC new_concept_name
‘Ophthalmologic treatment Right eye’ => ‘History of ophthalmologic treatment Right eye’
‘Premature gonadal failure [PhenX]’ => ‘History of premature gonadal failure [PhenX]’

Old concept names (values of a ‘long_common_name’ field in the source) will be added to the CONCEPT_SYNONYM.

please, see the file attached with a full list of such LOINC concepts.
new names for LOINC concepts indicating history.xlsx (52.0 KB)

@aostropolets, @Christian_Reich, @Alexdavv, @Alexdavv what do you think?


(Anna Ostropolets) #2

History of ophthalmologic treatment Left eye makes much more sense now. I’m assuming, there are more histories of in LOINC, but this list looks good.


(Alexander Davydov) #3

Currently, these concepts look the same like SNOMED historical concepts of clinical findings, but they are still questionnaires of LOINC vocabulary having predefined answers (yes/now/etc.).
Does it mean that we can use them the same way (without any answer as value) to indicate the fact of the presence of such a history?
Although the names are performing, I would say no, the original LOINC definitions tell us they’re just the questions.
To avoid any misunderstanding among users, can we express these concept names in the form of the questions? Or can we use some specific concept_class to isolate them?

Another option is to have a convention like: question without answer = question + yes. Is it in line with the CDM negative information convention and associated discussions?


(Polina Talapova) #4

Except concepts indicating a history with newly modified LOINC names, there are already existing ones with the similar name structure in the LOINC vocabulary (see the file: other LOINC concepts indicating history.xlsx )

And despite the fact LOINC and SNOMED concepts directly indicating “History of” seem to be similar, they are different from each other.
SNOEMD ‘History of’ concepts have a concept_class_id of ‘Context-dependent’ whereas LOINC ‘History of’ concepts have such concept_class_ids as ‘Clinical Observation’ and ‘Survey’.
SNOMED concepts can be used in mappings as either confirmatory concepts when observational source values are absent or indicative concepts for cases when observational source values exist while LOINC concepts always presuppose the existence of Answers, albeit not all such concepts have them currently.

Does it mean that we can use them the same way (without any answer as value) to indicate the fact of the presence of such a history? Although the names are performing, I would say no, the original LOINC definitions tell us they’re just the questions.

I also think that we shouldn’t use LOINC ‘History of’ concepts in the same way as SNOMED ones.

To avoid any misunderstanding among users, can we express these concept names in the form of the questions? Or can we use some specific concept_class to isolate them?

As I mentioned, they already have 2 specific isolating concept classes (‘Clinical Observation’ and ‘Survey’), which distinguish them from the SNOMED concepts. And I’m not sure that is obligatory to change their formulations to questions in this case.


t