Study Subject vs. Experimental Unit

The BRIDG model makes a distinction between a study subject and an experimental unit.  In most studies for which SDTM is implemented, these terms refer to the same person or animal, but there are studies where the study subject is different from the experimental unit. For those studies, it can be useful to understand these subtly different terms.

In the BRIDG model, a study subject is "A physical entity which is the primary unit of operational and/or administrative interest in a study."  In other words, the study subject is the entity that is enrolled in the study and given a study subject identifier.

In the BRIDG model, an experimental unit is "A physical entity which is the primary unit of interest in a specific research objective."

What's the difference?  Let's look at the kinds of physical entities that can act as experimental units.

  • A biologic entity (e.g., an animal or person)
  • A biologic entity part (e.g., an animal's or person's eye)
  • A biologic entity group (e.g., a group of animals or people)
  • A product (e.g., a medical device)
  • A product group
  • A specimen

Studies where a person or animal is both the study subject and the experimental unit (the primary unit of interest in a research objective) are the most common kinds of clinical and nonclinical trials.  In SDTM, USUBJID identifies the person or animal who is both the study subject and the experimental unit.

In SDTM, SPDEVID is used to identify a device, no matter what role the device plays; the device may be the study subject, an experimental unit, or involved only in some activity involving a human or animal study subject. If the study also involves people or animals who are units of "operational and/or administrative interest," those people or animals are identified using USUBJID. In some studies, devices are studied without involving people or animals.  In such studies, the device is the study subject and USUBJID will not be populated. 

In some studies, a person or animal is the study subject, but the experimental unit is a part of the person or animal.  Such studies occur, for example, in ophthalmology and dermatology. If parts of the person and animal (e.g., the eyes, or different areas of skin) are assigned to different treatments, the parts of the body are experimental units. In SDTM, USUBJID identifies with study subject (the person or animal) and FOCID identifies the experimental units within the study subject.

During the development of the BRIDG model, various examples were given in which a group of biologic entities would be experimental units.  These examples include studies where fluoride was added to water supplies or treatment was administered into tanks that contained multiple fish.  In the fluoride study example, the population to whom water was being supplied would be an experimental unit; in the second example, the group of fish in a tank would be an experimental unit.  The SDTM has a mechanism for recording observations about groups of biologic entities, (POOLID, defined using POOLDEF), but does not currently have a way to identify biologic entity groups as experimental units. The SENDIG-DART deals with the edge case where USUBJID could be considered to identify a pregnant animal and her fetuses. Whether a pregnant female is a biologic entity or a group of biologic entities is a subject for debate.

In the same way that a group of biologic entities can be an experimental unit, a group of products could be an experimental unit, although no examples spring to mind.

The BRIDG model is also applicable to in vitro experiments, where the experimental unit is a specimen.  In BRIDG terms, a "specimen" is not necessarily a specimen taken from a biologic entity.  For example, the BRIDG model could be used to describe a study in which specimens were samples of pharmaceutical product in a stability study, or a study in which specimens were organisms in a tissue culture.  The SDTM currently identifies biologic specimens using a sample identifier, a date, and either USUBJID or POOLID, since it assumes that a specimen was taken from a study subject or pool of study subjects. The SDTM does not currently have a way to handle specimens which are study subjects in their own right. If the SDTM is extended to deal with in vitro experiments, a mechanism will need to be developed.