We are proposing to use Group Resource when referencing a population or sample in Evidence Resources, in part to support re-use and interoperability across evidence, guidance, clinical decision support, and patient care.

In mapping ClinicalTrials.gov data (321 different elements in the schema and more than 370,000 records) to full expression in FHIR resources, we have some data where the "group" is a part of people, for example a study of 433 wounds in 276 patients.

One potential change is to allow more codes in the GroupType value set.  Another approach is add a new element (or extension) for Group.subpartType