RS relative-timing variables are no longer listed in RS

Rule

VAR-028

Category

Variables demoted

Basis

Draft Read from the public-review draft. The draft can still change before v4.0 is final, and this rule changes with it.

Detectability

External context required The Define-XML cannot settle this. Answering it needs a decision or a fact held outside the submission metadata.

Finding type

Question

Applies to

SDTMIG 3.2, 3.3, 3.4

What changes in v4.0

Four relative-timing variables — RSSTRTPT, RSSTTPT, RSENRTPT and RSENTPT — are present in the v3.4 Disease Response and Clin Classification specification table and absent from the v4.0 one. Timing variables are the case Section 3.2 addresses most directly: "any timing variables are permissible for use in any submission dataset based on a general observation class unless subject to a usage restriction in the SDTM". The roots survive across six to eight v4.0 domains apiece, and the RS domain assumption is consistent with it: RS permits timing variables generally — "any identifiers, timing variables, or Findings general observation class qualifiers may be added to the RS domain" — and the qualifiers it lists as ones that "would not generally be used" include no timing variable at all. This is the best-supported case in this section, because the general timing-variable guidance is a real permission for the category rather than an inference from silence. It still stops short of naming RSSTRTPT, RSSTTPT, RSENRTPT or RSENTPT, and no source explains why the RS table stopped listing them.

What to do about it

Do not migrate anything solely because these variables disappeared from the v4.0 RS specification table. Relative timing anchors are load-bearing for disease-response derivations, and both the general timing guidance and the RS assumption leave timing variables available to the domain. Note that basis in the reviewer's guide. If CDISC later publishes a restriction on these anchors, that becomes an analysis change requiring statistical review rather than a metadata edit — but nothing available today says so.

What the scan can and cannot see

The four anchors being declared is conclusive; that they must go is not claimed and is not supported. Section 3.2 states timing variables are permissible in any general-observation-class dataset unless restricted, the RS assumption lists no timing variable as an exception, and the restrictions that could change that are unpublished.