When Did That Happen? A Brief Guide to Representing Timing in SDTM.

In ordinary conversation, depending on what “that” is, the question, “When did that happen?” could be answered in many ways. The fact that there are so many ways to say when something happened helps to explain why there are so many timing variables in SDTM.

On June 16, 2015.

This answer suggests that whatever happened was of short enough duration that a separate start and end are not needed.  In SDTM, we generally expect to have a date or date/time (--DTC) for a finding.

It started on April 13 and lasted until April 20.

SDTM expects interventions and events may have longer durations, so it provides both start and end date or date/time (--STDTC and --ENDTC).

On the fourth day of our vacation.

This answer provides timing by counting from the start of a time period (a vacation) and is like expressing time relative to the start of a clinical trial using study days ( --DY, --STDY, and –ENDY).

It started while I was in college.

This answer, which expresses timing relative to a time period (while in college) is like the SDTM variables --STRF and --ENRF, which categorize the start and end of an event or intervention relative to the study reference period. The start and end of the study reference period are RFSTDTC and RFENDTC in the Demographics domain. The categories used are BEFORE, DURING, AFTER, and DURING/AFTER the study reference period.

Before April 15, 2015.

This answer also expresses timing by reference, but relative to point in time, rather than a period of time. In SDTM, the starts and ends of events and interventions can be represented using --STTPT and --ENTPT for the point in time and –STRTPT and –ENRTPT for the relative time category i.e., BEFORE, COINCIDENT WITH, ONGOING, or AFTER. The point in time can be represented either as a date/time or as a named time in the study, such as a visit.

At the fourth tee.

This answer is like a study visit or time point. In golf, there’s a plan for progressing through the game, and the player is expected to progress around the course, going to the various tees in order. In a study, the subject is expected to participate in a sequence of visits. SDTM has three variables used to describe visits (VISIT, VISITNUM, and VISITDY) and four used to describe time points (TPTNUM, TPT, ELTM, TPTREF).

In the bottom of the fifth inning.

This answer is a little different from the golf answer, since the innings are periods of time, rather than points in time.  A study is divided into periods of time we call epochs and elements, represented in SDTM using the variables EPOCH and TAETORD.

Our offer was accepted on May 20, we closed on May 31, and moved in on June 4.

This answer illustrates something (buying a house) whose timing could be defined in several ways. The starts of many diseases have similar ambiguities, which led to the introduction of the variable MHEVDTYP in SDTMIG v3.3. MHEVDTYP allows one to say, for instance, that symptoms started on one date, while diagnosis was on another date, and confirmation of the diagnosis by a particular test was on a third date.

Before I was married. At the time my son was born.

These answers describe timing relative to important events in a person’s life. In clinical trials, it is sometimes important to show, for each subject, a timeline of the course of the disease under study, and to relate other observations about that subject to the “milestones” on that disease timeline. This led to the “disease milestones” set of domains and variables introduced in SDTMIG v3.3.

When I was six years old.

The tenth answer provides timing relative to the person’s birth date. SDTM has only one age, the age in the Demographics domain, which is the subject’s age at the study reference start date, RFSTDTC. However, timing is sometimes collected as an age e.g., “age at diagnosis.”  At present there is no standard SDTM variable to hold the age of the subject at the time of an event, intervention, or finding.

Did X happen during Y time period?

SDTM has two timing variables, which only indirectly describe when something happened. These are --EVLINT and --EVLTXT, which describe an evaluation interval or recall period.  These variables really describe an aspect of a question, such as “Did X occur during the recall period?” or “How frequently did X occur during the recall period?” or “How severe was X during the recall period?” The evaluation interval provides a place to represent this aspect of the question. The answer to the question applies only to “X” during the evaluation period. The Findings About (FA) structure provides a place, the --OBJ variable, to represent the “X” part of the example questions above.