Open ASL-rmarshall opened 3 months ago
If it's true that a population must always be specified for a study design, do we need a USDM v3.0 rule to enforce this requirement?
Note that the rule created for check CHK0190 (DDF00050) - study arms must only reference study cohorts that are defined within the population defined within the same study design as the study arm - relies on a population being defined for the study design.
Currently (as of release 3.4), the StudyDesign.population
relationship is a "definitional" relationship in the API specification - the definition of the StudyDesignPopulation
is included within the StudyDesign
; the StudyDesignPopulation
is not referenced by id. Similarly, the StudyDesignPopulation.cohorts
relationship is also "definitional" - study cohorts can only be defined within the context of a StudyDesignPopulation
.
What happens if you have 2 study designs that use different cohorts of the same population?
At the moment, you'd have to duplicate the population definition in the 2 study designs so that the study-design-specific cohort definitions can be defined within the relevant study design. If the population/cohort definitions were moved to a (n API only) collection under StudyVersion
(in the same way that EligibilityCriterion
definitions have been moved), it would be possible to create a single population definition (containing definitions for all the cohorts that are needed) and then to reference the population and relevant cohorts by id.
The cardinality for the
StudyDesign.population
relationship is currently 0..1, which implies that specifying a population for the study design is optional. I think the cardinality should be changed to 1 to indicate that a population must always be specified for a study design.