Our recently held webinar Confusing Validation Rules Explained sparked lots of follow-up questions from you. In this post we will clarify the guidance on populating the EPOCH variable and give you two effective ways to derive EPOCH values from other data when direct EPOCH values are not available. Refer here for more Confusing Validation Rules Q&A.
What is the EPOCH Variable
The EPOCH variable specifies what phase of the study data corresponds with. Including accurate EPOCH data allows the reviewer to easily determine which phase of the study the observation or the event occurred, as well as the intervention the subject experienced during that phase.
EPOCH is Expected
From the FDA Study Data Technical Conformance Guide: "The variable EPOCH should be included for clinical subject-level observation (e.g., adverse events, laboratory, concomitant medications, exposure, and vital signs)."
Do not make the mistake of treating EPOCH as optional because CDISC guidance lists EPOCH as a permissible variable. For the FDA, EPOCH is expected. The same is true for the PMDA.
Handling Missing Data Used to Derive EPOCH
When the EPOCH variable is not populated directly, another timing variable, such as VISIT, may serve as an alternative source for EPOCH data. EPOCH may also be determined from the schedule of assessment for when specific data is captured, even if a date is missing.
Often, partial dates can be used to derive EPOCH by comparing the date components provided in the partial date to those components in the date is used to derive EPOCH. For example, if the observation or start date is 2020-05, that may be enough to determine the EPOCH. Make sure the derivation accounts for these partial dates. A partial date may not be sufficient to determine EPOCH.
If you cannot populate EPOCH from the available information, then leave EPOCH null and explain the situation in the reviewer's guide.
Handling Errors in Data Used to Derive EPOCH
Work with your data management group to correct data errors that affect EPOCH derivation, such as errors with dates. If correction is not possible, explain the situation in the reviewer's guide.
If you have EPOCH values that don't have a corresponding match in CDISC CT, explain that issue in the reviewer's guide. List the values that you are using to extend the codelist. Also, submit the additional terms to CDISC for adding to CDISC CT.
Correct EPOCH Position in SE Domain
EPOCH should come before SESTDTC in the domain.
This relates to the difference between variable order listed in the SDTMIG v3.2 and SDTM v1.4. This difference was corrected in the SDTMIG v3.2 Errata, which states "TAETORD and EPOCH should come before SESTDTC in the table". You should follow the corrected guidance from the SDTMIG v3.2 Errata.
How to Derive EPOCH in the DS Domain
SDTMIG v3.3 guidance is: "In DS, EPOCH is based on DSSTDTC." Therefore, starting in SDTMIG v3.3, EPOCH is derived in a consistent way with the other SDTM domains.
Historically, in SDTMIG v3.2 and earlier, guidance was: "When EPOCH is populated for disposition events (records with DSCAT = DISPOSITION EVENT), EPOCH is the name of the Epoch for the disposition event described in the record. This is a subtly different meaning from that of EPOCH when it is used in other general-observation-class domains, where EPOCH, as a Timing variable, is the name of the Epoch during which --STDTC or --DTC falls.
For the webinar video, slides and FAQs, refer to the original blog post: Confusing Validation Rules Explained.