Open mikix opened 2 months ago
we talked about this off thread, and the preferred approach to this, and more generally for extensions that are well defined, is to create a column for each possible field - this allows for maximal downstream flexibility.
so in this case we would have race_ombSystem_code
, race_ombStytem_display
(and perhaps a race_ombSystem_version
and race_ombSystem_userSelected
) from the ValueCoding, a similar set for race_detailed_*
, and a race_text
from the ValueString.
The race and ethnicity extensions offer several levels of detail:
ombCategory
codings (10,000 foot view),detailed
codings (nitty gritty), andtext
string (:shrug:).We currently only examine
ombCategory
anddetailed
codings (preferringombCategory
values). And only actually use thedisplay
values found there.Should we also examine
text.valueString
?text
is actually the only required field and we only use display values anyway. But... who knows what the data quality is like there.A part of me doesn't even like looking at
display
values, instead preferring to stick to theombCategory
values, looking at the code/system, and providing the string ourselves. That way, we don't rely on any weird EHR presentations of the strings. But maybe that's too paranoid. (Note: I did get that paranoid though for the data-metrics study)