Variable names suggest adjudication content the dataset label does not

Rule

DOM-008

Category

New domains

Basis

Confirmed Traceable to the published specification text cited below.

Detectability

Define-XML heuristic The Define-XML indicates this, but does not prove it. Treat the finding as a prompt to check, not as a verdict.

Finding type

Heuristic

Applies to

SDTMIG 3.2, 3.3, 3.4

What changes in v4.0

v4.0 introduces EA (Event Adjudication), a Findings About domain for adjudication results, with EAOBJ naming the object of the adjudication and EATESTCD, EATEST, EAORRES and EASTRESC carrying the assessment. Adjudication content in a legacy study is rarely in a dataset called anything obvious: it is commonly a handful of qualifier columns on the domain being adjudicated. DOM-002 finds the case where the DATASET is labelled for adjudication; this rule finds the case where the dataset label says nothing and a variable label does.

What to do about it

Confirm first that this really is adjudication content rather than a variable that mentions adjudication — the match is a label, not a diagnosis. If it is, EA is where v4.0 would put it, and moving it there is optional for a legacy study: nothing requires the remap. What it buys is that a reviewer looking for adjudication results finds them in the domain designed for them, with the committee and evaluator metadata EAEVAL and EAEVALID carry, rather than in a qualifier column on another domain.

What the scan can and cannot see

Matched on a VARIABLE label containing "adjudicat" in a dataset whose own label does not. A label is a proxy for content: it misses adjudication data labelled anything else, and it matches a variable that merely mentions adjudication in passing. What the column holds is record-level and is not claimed.