What is OHDSI’s approach to post coordination?
For oncology, metastasis and secondary cancers are mapped to standard concepts in the Measurement domain. Will this be done for other conditions?
Other conditions? Not following. They don’t metastasize. Or what do you have in mind?
Conditions which only occur during/after another condition.
Some examples off the top of my head: intrahepatic cholestasis during pregnancy, right heart failure due to pulmonary hypertension, diabetic ketoacidosis,
I see what you are thinking. The idea is not bad, but has a flaw: All those conditions that can occur during/after another one (and is etiologically related) can also occur alone. While the metastasis can only occur as a result of a primary cancer. Which means, your right heart failure will show up sometimes in the Condition and sometimes in the Measurement domain. In addition, the logical combination is only putative: You can have a right heart condition before the pulmonary hypertension.
What are you trying to achieve? What is the use case?
Thanks, @Christian_Reich,
The use case is educating those new to OHDSI and OMOP CDM. It’s easy for me to explain why ICD10CM = R79.82 , elevated C-reactive protein is in the Measurement domain. It’s a measurement of C-reactive protein in the blood. I didn’t have an answer to why cancer was in the Measurement domain since cancer is a Condition, not a Measurement
Ah! Yes, that does require an explanation and justification, and it is a somewhat ugly compromise:
When we started the Oncology WG a while ago (it’s slowly getting 10 years), we faced the problem that cancer diagnoses are way more detailed and complicated than the other diseases. Technically, all those “Stage 4 squamous cell carcinoma of the upper left lobe of the lung with BRAF V600E mutation” should be precoordinated Condition concepts. The problem is that such precoordination would have resulted in permutational explosion due to the sheer number of combinations of its components (attributes, cancer modifiers). In SNOMED alone, there are:
So, precoordination was not an option. We had to postcoordinate them.
But there is another reason: Many of them are dynamic, i.e. would come and go. A metastasis can be detected, but then resolve from successful treatment. So, we also needed a value.
The only places to put these were Measurement or Observation. Since many of them are the result of distinct testing during the diagnostic workup, that’s where they fell.
We realize this is a little bit of a Frankenstein’s monster. But it works well. We are going to publish a good guidance document very soon.
2 Likes