​The following study variable summary highlights the basic format one can expect when producing as well as consuming these summaries. The code is set to a common code, C0242482 which indicates that the Observation is a Summary Report. The subject property is set the the population from which the summary data was collected, in this case the Entire Study Population. The Focus references the study itself. For the value[X] portion of the resource, we have described the source of the summary data which happens to be the “gender” code from the demographics-table along with a harmonized coding mapping that variable to the UMLS code for “Gender”, C0079399.

Finally, we have the summary data listed as components.

These are enumerated alongside a corresponding code which indicates what the summary component is about. In this particular case, it’s codes for male and female from administrative gender and for missing, which is coded as the UMLS code, “C1705492”, “Missing”.

Generated Narrative: Observation

status: final

code: Summary Report (umls#C0242482)

subject: Group/example-summary-group-1

focus: ResearchStudy/ncpi-research-study-01

value: Gender (umls#C0079399; Datatable CodeSystem (lists all variables)#gender)

component

code: Female (AdministrativeGender#female)

value: 205

component

code: Male (AdministrativeGender#male)

value: 183

component

code: Missing (umls#C1705492)

value: 12

This data does not reveal details about any particular subject, and as such can be hosted on public FHIR server regardless of the restrictions put into place on the source data from which it was derived. By tying the valueCodeableConcept to public ontologies, the data becomes more easily discovered. Finally, if the values being enumerated are not common terms, like the ones shown above, if those too rest side-by-side with public terms, the data is far more easily interpreted by other researchers.