SAS tip: keep partial dates and imputation separate

Preserve the original value

Retain the collected date string and represent what is actually known. Do not silently turn a year-only value into a complete date. Derive analysis dates under the agreed specification, with imputation flags where required.

Test the boundary cases

Include leap years, incomplete months, dates before treatment and impossible date strings. Assess whether a chosen imputation changes baseline status or treatment-emergent classification. Escalate ambiguous rules to the responsible statistical lead.

Sources & further reading

• CDISC — ADaM

Sources checked 20 September 2026. Implementation commentary reflects a practical review perspective; source material may change.