The short version
- In most of the US, clocks went back an hour at 2:00 this morning, so local times between 1:00 and 2:00 happened twice.
- SDTM stores dates and times as ISO 8601 text and keeps partial dates partial; imputation belongs in ADaM, with flags.
- Use this week to check three things: how times are collected, how partial dates are imputed, and how elapsed times are derived.
Why a time-zone change is a data question
ISO 8601 defines unambiguous text representations of dates, times and durations, including an optional offset from UTC [1]. SDTM uses ISO 8601 for its date and time variables, the --DTC variables, and represents a partial date or time by omitting the unknown components from the right, so "2026-03" is a valid date for something known only to the month [2]. Clinical systems usually record local clock time without an offset. On most days that is harmless. On a night like last night, a time such as 01:30 could mean two different moments an hour apart, and in March one local hour does not exist at all.
For most endpoints this does not matter. For anything where elapsed time between events is short and precise, such as pharmacokinetic sampling, dosing intervals or timed ECGs, an hour's ambiguity can change a derived value.
Three checks for the week
1. How are times collected and stored? Ask data management whether the EDC, the PK system and any device data record local time, UTC or local time with an offset, and whether the sources agree. Mixed conventions are the real risk: a device that stores UTC and an EDC that stores local time will disagree by several hours, and by a different amount on each side of a clock change.
2. How are partial dates imputed? SDTM keeps the partial date. ADaM is where imputation happens, in analysis date and time variables, with imputation flags that record what was imputed, for example the day, or the time [3]. Check that the imputation rules are written in the SAP or the analysis specifications, that the flags are populated and that the rules are the same across datasets. Adverse event start dates and concomitant medication dates are the usual places where imputation differs between programmers.
3. How are elapsed times derived? Study day, time since dose and duration of exposure are derived from two dates or date-times. Confirm which reference date is used, how a missing time is handled and what happens across a clock change. For PK and timed assessments, ask whether the planned and actual times are available and whether derivations use actual times.
Partial dates in practice
| Collected | SDTM --DTC | Typical ADaM handling |
|---|---|---|
| 14 March 2026 at 09:30 | 2026-03-14T09:30 | Used as is |
| 14 March 2026, no time | 2026-03-14 | Date used; time not imputed unless the analysis needs it |
| March 2026 | 2026-03 | Day imputed by the rule in the SAP, with the imputation flag set to day |
| 2026 only | 2026 | Month and day imputed by rule, or left missing, as the SAP states |
A synthetic example
In the fictional ZEPHYR-101 PK substudy, sample times are entered as local clock time at its US East Coast sites, while dosing comes from an electronic diary that stores UTC. The derived time since dose is out by five hours for every sample in winter and four in summer, but the plausibility checks pass because the values are positive and in range. Converting both sources to a common basis before derivation, and documenting it in the reviewer's guide, fixes every record at once.
Do this today
- Ask which time basis each data source uses for dates and times: local, UTC or local with offset.
- Compare the imputation rules for partial adverse event and medication dates across your analysis datasets.
- Pick one subject with dosing and sampling on a clock-change weekend and check the derived elapsed times by hand.
Check yourself
One question
Show the answer
Answer 2. SDTM represents dates as ISO 8601 text, and a partial date is shown by leaving out the unknown components from the right. Any imputation, such as using the first of the month, belongs in the analysis data, where it is flagged and documented.
Apply it
Turn it into a plan
The brief says what to check. The Navigator and the Handbook turn it into dated, owned actions for your milestone, in your browser, with no study data.
Sources
References
- ISO. ISO 8601-1:2019 Date and time: Representations for information interchange, Part 1: Basic rules. International Organization for Standardization; 2019. www.iso.org ↗
- CDISC. Study Data Tabulation Model Implementation Guide (SDTMIG) version 3.4. www.cdisc.org ↗
- CDISC. Analysis Data Model Implementation Guide (ADaMIG) version 1.3. www.cdisc.org ↗
Standards and guidance change. Confirm the version in force for your study, sponsor and region before relying on any reference. Examples are synthetic and illustrative. This brief is generalized professional guidance, not regulatory advice.