Need for archetypable flavours of null to be specified on a per archetype basis in addition to the current flavours of null, as per discussions with Sam. These flavours of null will likely need atcodes.
I agree that there is a requirement, but it only relates to a relatively small sample of clinical data; the vast majority of clinical data have no need of any 'reason for null' item. For that reason, I think adding it to the RM doesn't make sense.
If we think about it, it's clear that the possible reasons for null are directly related to the kind of content. E.g. ED admission, where patient can be unconscious etc. I think it would make more sense to simply model this information as part of the relevant archetypes, probably for COMPOSITIONs for ED and other similar situations.
Null flavours themselves only relate to a small sample of clinical data, but it's not uncommon to have to specify why information is missing, particularly in reporting contexts. With several other issues, such as what we've been calling "exclusion of tests/scorings/observations", we've been told "this should be modeled using null flavours". That would be great, but it's not possible without being able to specify the reason for null.
I would reiterate my support for having 'reason for null'. Having this substantially reduces the complexity of having to potentially model some sort of comment or annotation against every data point.
I seem to remember this was the conclusion we got to last time we discussed this. This avoids many types of complexity.
At the SEC meeting we decided that the easiest way to support this was via as part of RM 1.1.0 So any element can have an optional ‘Reason for null’ applied - it is a DV_TEXT so could be made into an internal code list and constrained further in archetypes or templates.