cgreich
commented
4 years ago @MarkWlodarski et al.:
We agree, distinguishing between the disposition for an allergy (Allergy to food) and an actual reaction after intake (Allergic reaction to food makes sense. In the SNOMED hierarchy, the actual foods are added as attributes in the descendants of these concepts. It is true that the combos aren't enough, given the many foods that are known to call allergies. I am not sure what "Observed food reaction" would add, since it is similar to the latter of the two above, as you also stated.
We believe we have found a simple workflow using existing OMOP concepts that can model this case. In the case of a new food reaction it will be simple enought to link the Generic Food Reaction concept to a food stored in the Observation table using the "caused by" link in the fact_relationship table. In the case that the food itself is unmodeled, we would be able to use a generic "food" in the observation table and then add the new food as a string.
Don't do that @MarkWlodarski. Here is why: "Finding workflows" is not a good thing, even if it sounds like it. Remember: OMOP is a standard. It is actually not so useful for you, but for other people who do not have access to your data, but still can query. And for building tools and methods. So, if the current Condition hierarchy keeps "Allergy to chocolate" as a descendant of "Allergy to food" than folks will expect that to be the only mechanism. They won't look for some Fact Relationships and do those rather fancy joining acrobatics. And certainly no strings: We are fully normalized to the CONCEPT reference table in the OMOP Standardized Vocabularies.
Why don't we do what we said we would do: We add SNOMED Extension concepts for those food allergies that SNOMED misses. We also petition to them to add them, but that will take time, as you correctly pointed out. Till then, we will have them in the OMOP Vocabularies and in the right place in the hierarchy as children of "Allergy to food" and "Allergic reaction to food". Happy to do that.
Do you need all the foods in your spreadsheet combined with either one?
MarkWlodarski
Alexdavv
wlodarmt
OMOP Team,
Thank you again for taking the time to talk to us on the 24th about the changes we were proposing based on the data dictionary we have created for Food Allergies.
Following our internal discussion we are looking for the following changes to be made to OMOP:
1) Allergy to Food - Concept ID 4188027 and its associated children should be moved to the Condition domain in order to reflect their status as a clinical diagnosis of an existing condition. This is under the advice of the panel of Food Allergy experts we have consulted during the creation of our Food Allergy Data Dictionary.
2) The children of Allergy to Food should be expanded to include the complete list of Food Allergens identified by our experts, I have attached this list. FoodAllergiesOMOP.xlsx
During our call on the 24th of July we had orignally proposed the creation of a new concept "observed food reaction" to allow us to model newly discovered or previously unrecorded reactions to foods. However we are no longer pushing this for two reasons: Primarily, it will be difficult for SNOMED to be convinced to add this new concept, as it is not significantly different than the existing concepts "Allergy to Food" or "Allergic Reaction to Food".
Second, we believe we have found a simple workflow using existing OMOP concepts that can model this case. In the case of a new food reaction it will be simple enought to link the Generic Food Reaction concept to a food stored in the Observation table using the "caused by" link in the fact_relationship table. In the case that the food itself is unmodeled, we would be able to use a generic "food" in the observation table and then add the new food as a string.
Thank you once again, The Northwestern University Food Allergy Data Commons Team