We looked at another example of percentage to consider both the counts per 100 cells (number fraction) and a percentage fraction of substance in a fluid (quantity fraction) Fractions definition - LOINC has been consistent to say some portion of whole e.g. basophils in white cells New lab techniques are not manual counts of cells on slides. Raw fraction vs percent can be distinguished by the units. LOINC will only differentiate codes by property, not the same kind of property with different units. You will have to map codes that do differentiate to the same LOINC code - or in the same ontology they will be more specific children of the non-specific code In NPU these, percent and per mil, are differentiated. This is consistent with UK approach to differentiate SNOMED CT concepts and supports absolute counts in a volume of fluid which is important in clinical practice to represent The only way to represent these currently in SNOMED CT as fully defined is by including units as percent Awaiting further internal discussions in the LOINC Ontology design space - currently they say the number fraction cell counts are outdated. If they do per 100 leukocytes as percent, then they are likely to do per 1000 erythrocytes as number fraction SNOMED CT policy - If the concept says percent then it should be modelled with the unit, otherwise if it says 'fraction' it does not need the unit - these are the rules that apply in authoring SNOMED CT generally Contravenes editorial policy to say 'percent' in the SNOMED CT description and not model percent unit if fully defined / equivalent class axiom e.g. international content for cancer synoptic reporting by immunohistochemistry 1255078008 | Percent of cell nuclei positive for proliferation marker protein Ki-67 in primary malignant neoplasm by immunohistochemistry (observable entity) | There are questions around how do we standardise with the other lab subject matter groups developing content independently that is inconsistent? We should be applying the same editorial design principles and SNOMED CT authoring policy Further inconsistencies in the cancer synoptic reporting concepts e.g. 1 - modelling time_aspect -> single point in time but not stating this in the lexical description In contrast to LOINC Ontology examples that do state 'at single point in time' in the description and model in the logic e.g.2 - use inheres_in whereas LOINC Ontology used direct_site Doesn’t specify specimen in the text but then neither do UK PaLM and LOINC - logical definitions are more elaborately specify 'specimens' |