Why are ImagingStudies dependent on Conditions, and can I change that?

[Gilgeam]

Greetings friends,

I've been building a model for radiological follow-ups on imaging pathologies and run across a design choice (?) that I found thoroughly confusing: I had this model where a population would get Imaging and get diagnosed an unclear mass, receivied a ConditionOnset for unclear mass. Then I had a series of follow ups, which led to this mass being identified as benign or malign. I coded this as the Condition of "Unclear mass" ending and another COndition of "Benign (or malign) mass" starting.

However, when I end the Condition, the corresponding ImagingStudy gets wiped out from the FHIR datasets too. I end up with just the data from the last examination, which has the information that its benign. So, for instance, in a workflow of CT -> MRI, the fact that I did the CT gets deleted from the record too.

I found this thoroughly unintutive, but from the testing I did, this seems to be the case. Can I somehow avoid this or do I need to preserve all the previous conditions? ALso, is that behavior documented someowhere that I just missed?

Thanks a lot for any clarification, it would really be appreciated!

[hadleynet]

Looking at the code I don't see any obvious dependency from imaging study on condition so I'm puzzled by the behavior you are seeing. Are you wrapping the imaging study in an encounter?

[Gilgeam]

Update: I'll leave my former reply below, but @jawalonoski pointed me in the direction of the mistake - the model behaves fine, the old Condition along with that ImagingStudy are likely just not categorized as important enough to not get filtered out based on whatever priority has been decided for the filter process. If I disable the export filter, both the Imaging Study and the old condition are in the dataset.

---

Yes. My original model is fairly complex, but I made a custom testing one to figure out what was going on. I'll try attaching the .jsons below, built with the Module Builder.

Essentially, its constructed as

Initial -> Delay -> ConditionOnset (Condition 1, at Encounter 1) -> Encounter 1 -> ImagingStudy CT -> EncounterEnd -> Delay -> Encounter_MRI -> ImagingStudy MRI -> ConditionOnset (Condition 2) -> EncounterEnd.

If I keep this like this, I have both the CT and MRI in my dataset. But if I modifiy it as

..... Delay -> Encounter_MRI -> ImagingStudy MRI -> ConditionEnd Condition1 -> ConditionOnset Condition2 -> EncounterEnd

... then the CT disappears and my FHIR data only contains the MRI.

I repeated this test a lot of times, but the behavior stays consistent. I'd be curious if its the same with you.

Condition_Not_Ended.json
Condition_Ended.json

[jawalonoski]

If you're missing data, it could be due to the years of history setting. See why isn't X showing up? and Common Configuration - Other Export Settings

[Gilgeam]

YES! Progress! Thank you very much!

I didn't expect this to be a thing, but I tried disabling the export filter in the setting so everything gets exporter and this means the data that was previously scrapped now shows up! That's great! That means I can now end those conditions!

I am still confused why the Imaging Study would get filtered out along with the ended condition. I suspect that the logic is that not all studys are relevant long term and will only remain if somehow connected to an ongoing condition?!