Technology preview and application of co0ncept model
Review and discuss editorial advice for application of concept model in creation of observables content
Comments
Looking at the concept, it appears to be measuring the drug 'Morphine' not the substance....but it sounds like the query you want is SELECT * WHERE COMPONENT HAS_ACTIVE_INGREDIANT Morphine (substance). Would this query capture all of the observables containing Morphine e.g. free Morphine and others? Probably need better guidance, but it comes down to exactly what the observation is measuring, and in this case it appears to be the drug morphine.
The product concepts in SNOMED CT are defined by substance concepts via the “Has active ingredient (attribute).” This relationship can be used in queries if needed, as Michael also points out. Yes, it is an additional step, but it keeps the Observables model “cleaner,” more concise and consistent.
Even in the measurement of the drugs (or any product), what is actually being measured is the active ingredient. So while the range of “Component (attribute)” includes both products and substances, we would still go with substances in the vast majority of cases.
If the LOINC Term states “morphine,” we cannot assume or model anything more specific than the LOINC Term (unless RII clarifies that a different substance is actually measured).
We also want to add that the Product hierarchy in SNOMED CT is currently undergoing changes. One change is related to making FSNs more clear, for example, 73572009 | Morphine (product) will change to 73572009 | Product containing morphine (medicinal product) for the July 2017 release. Another change is that all product concepts will be fully defined with an “Has active ingredient (attribute)” relationship to a substance concept.
We realize we need to update the modeling guidelines in the documentation for the LOINC - SNOMED CT Cooperation Project release files in this area.
If more discussion or clarification is needed, we can bring this topic to the next Observables call.
An unfortunate feature of the concept model is that drugs have conceptual content in both substance and pharmaceutical products which are a single 'thing' from the standpoint of the clinical world of users. In developing some groupers (query applications) for the observables tech preview, I observe that we are placing our users in the difficult position of writing two queries to answer simple questions like: 'Did the patient have a test for opioids?' Reviewing the MRCM for observables, I note that 'Component' and 'Inheres in' are cardinality 0...* AND have valuesets of both Substances and Pharmaceutical products. To favor efficient use of observable content for query and decision support purposes, should we develop guidance that observables assessing 'presence' or 'process' relating to chemical compounds which are also pharmaceutically active should always be defined with BOTH attribute value pairs, assuring that the naive user will get proper performance of database queries like "Was there a test for opioids?'. So, the tech preview model would change for 'Morphine level in blood', to wit:
Morphine Mass/volume in serum or plasma (observable entity)
fully defined as
IS_A Opitaes Mass/volume in serum or plasma
COMPONENT Morphine(substance)
COMPONENT Morphine(Pharmaceutical product)
DIRECT SITE Acellular blood specimen
INHERES IN Plasma (substance)
PROPERTY TYPE Mass concentration
SCALE TYPE Quantitative
TIME ASPECT Point in time