You are viewing an old version of this page. View the current version.

Compare with Current View Page History

« Previous Version 3 Next »


  1. Investigator and site identification: Companies use different methods to distinguish sites and investigators. CDISC assumes that SITEID will always be present, with INVID and INVNAM used as necessary. 
  2. Every subject in a study must have a subject identifier (SUBJID). In some cases a subject may participate in more than 1 study. To identify a subject uniquely across all studies for all applications or submissions involving the product, a unique identifier (USUBJID) must be included in all datasets. Subjects occasionally change sites during the course of a clinical trial. Applicants must decide how to populate variables such as USUBJID, SUBJID and SITEID based on their operational and analysis needs, but only 1 DM record should be submitted for each subject. The Supplemental Qualifiers dataset may be used if appropriate to provide additional information.
  3. Concerns for subject privacy suggest caution regarding the collection of variables like BRTHDTC. This variable is included in the Demographics model in the event that an applicant intends to submit it; however, applicants should follow regulatory guidelines and guidance as appropriate.
  4. With the exception of trials that use multistage processes to assign subjects to arms described below, ARM and ACTARM must be populated with ARM values from the Trial Arms (TA) dataset and ARMCD and ACTARMCD must be populated with ARMCD values from the TA dataset or be null. The ARM and ARMCD values in the TA dataset have a one-to-one relationship, and that one-to-one relationship must be preserved in the values used to populate ARM and ARMCD in DM, and to populate the values of ACTARM and ACTARMCD in DM.
    1. Rules for the arm-related variables: 
      1. If ARMCD is null, then ARM must be null and ARMNRS must be populated with the reason ARMCD is null.
      2. If ACTARMCD is null, then ACTARM must be null and ARMNRS must be populated with the reason ACTARMCD is null. Both ARMCD and ACTARMCD will be null for subjects who were not assigned to treatment. The same reason will provide the reason that both are null.
      3. ARMNRS may not be populated if both ARMCD and ACTARMCD are populated. ARMCD and ACTARMCD will be populated if the subject was assigned to an arm and received treatment consistent with 1 of the arms in the TA dataset. If ARMCD and ACTARMCD are not the same, that is sufficient to explain the situation; ARMNRS should not be populated.
      4. If ARMNRS is populated with "UNPLANNED TREATMENT", ACTARMUD should be populated with a description of the unplanned treatment received.
    2. Multistage assignment to treatment: Some trials use a multistage process for assigning a subject to an arm (see Section 7.2.1, Trial Arms, Example Trial 3). In such a case, best practice is to create ARMCD values composed of codes representing the results of the multiple stages of the treatment assignment process. If a subject is partially assigned, then truncated codes representing the stages completed can be used in ARMCD, and similar truncated codes can be used in ACTARMCD. The descriptions used to populate ARM and ACTARM should be similarly truncated, and the one-to-one relationship between these truncated codes should be maintained for all affected subjects in the trial. Example 3 below provides an example of this situation; see also Section 5.3, Subject Elements, Example 2. Note that this use of values not in the TA dataset is allowable only for trials with multistage assignment to arms and to subjects in those trials who do not complete all stages of the assignment.
    3. Examples illustrating the arm-related variables
      1. Example 1 below shows how to handle a subject who was a screen failure and was never treated.
      2. The Subject Elements (SE) dataset records the series of elements a subject passed through in the course of a trial, and these determine the value of ACTARMCD. The following examples include sample data for both datasets to illustrate this relationship.
        1. Example 2 below shows how subjects who started the trial but were never assigned to an arm would be handled.
        2. Section 5.3, Subject Elements, Example 1 illustrates a situation for a subject who received a treatment that was not the one to which they were assigned.
        3. Section 5.3, Subject Elements, Example 2 illustrates a situation in which a subject received a set of treatments different from that for any of the planned arms.
  5. Study population flags should not be included in SDTM data. The standard supplemental qualifiers included in previous versions of the SDTMIG (COMPLT, FULLSET, ITT, PPROT, SAFETY) should not be used. Note: The ADaM Subject-level Analysis Dataset (ADSL) specifies standard variable names for the most common populations and requires the inclusion of these flags when necessary for analysis; consult the ADaMIG for more information about these variables.
  6. Submission of multiple race responses should be represented in the Demographics (DM) domain and Supplemental Qualifiers (SUPPDM) dataset as described in Section 4.2.8.3, Multiple Values for a Non-result Qualifier Variable. If multiple races are collected, then the value of RACE should be “MULTIPLE” and the additional information will be included in the Supplemental Qualifiers dataset. Controlled terminology for RACE should be used in both DM and SUPPDM so that consistent values are available for summaries regardless of whether the data are found in a column or row. If multiple races were collected and 1 was designated as primary, RACE in DM should be the primary race and additional races should be reported in SUPPDM. When additional free-text information is reported about subject's race using “Other, Specify”, sponsors should refer to Section 4.2.7.1, "Specify" Values for Non-Result Qualifier Variables. If race was collected via an "Other, Specify" field and the sponsor chooses not to map the value as described in the current FDA guidance (see CDISC Notes for RACE in the domain specification), then the value of RACE should be “OTHER”. For subjects who refuse to provide or do not know their race information, the value of RACE could be “UNKNOWN”. See DM Example 4, DM Example 5, DM Example 6, and DM Example 7.
    1. The Racec-Ethnicc Codetable (available at https://www.cdisc.org/standards/terminology/controlled-terminology) represents associations between collected race values and published race Controlled Terminology, as well as collected ethnicity values and published ethnicity Controlled Terminology. 
  7. RFSTDTC, RFENDTC, RFXSTDTC, RFXENDTC, RFCSTDTC, RFCENDTC, RFICDTC, RFPENDTC, DTHDTC, and BRTHDTC represent date/time values, but they are considered to have a record qualifier role in DM. They are not considered to be timing variables because they are not intended for use in the general observation classes.
  8. Additional permissible identifier, qualifier, and timing variables:
    1. Only the following timing variables are permissible and may be added as appropriate: VISITNUM, VISIT, VISITDY. The record qualifier DMXFN (External File Name) is the only additional qualifier variable that may be added, which is adopted from the Findings general observation class, may also be used to refer to an external file, such as a patient narrative.
    2. The order of these additional variables within the domain should follow the rules as described in Section 4.1.4, Order of the Variables, and the order described in Section 4.2, General Variable Assumptions.
  9. As described in Section 4.1.4, Order of the Variables, RFSTDTC is used to calculate study day variables. RFSTDTC is usually defined as the date/time when a subject was first exposed to study drug. This definition applies for most interventional studies, when the start of treatment is the natural and preferred starting point for study day variables and thus the logical value for RFSTDTC. In such studies, when data are submitted for subjects who are ineligible for treatment (e.g., screen failures with ARMNRS = "SCREEN FAILURE"), subjects who were enrolled but not assigned to an arm (e.g., ARMNRS = "NOT ASSIGNED"), or subjects who were randomized but not treated (e.g., ARMNRS = "NOT TREATED"), RFSTDTC will be null. For studies with designs that include a substantial portion of subjects who are not expected to be treated, a different protocol milestone may be chosen as the starting point for study day variables. Some examples include non-interventional or observational studies, studies with a no-treatment arm, and studies where there is a delay between randomization and treatment.
  10. The DM domain contains several pairs of reference period variables: RFSTDTC and RFENDTC, RFXSTDTC and RFXENDTC, RFCSTDTC and RFCENDTC, and RFICDTC and RFPENDTC. There are 4 sets of reference variables to accommodate distinct reference-period definitions and there are instances when the values of the variables may be exactly the same, particularly with RFSTDTC-RFENDTC and RFXSTDTC-RFXENDTC. 
    1. RFSTDTC and RFENDTC: This pair of variables is sponsor-defined, but usually represents the date/time of first and last study exposure. However, there are certain study designs where the start of the reference period is defined differently, such as studies that have a washout period before randomization or have a medical procedure required during screening (e.g., biopsy). In these cases, RFSTDTC may be the enrollment date, which is prior to first dose. Because study day values are calculated using RFSTDTC, in this case study days would not be based on the date of first dose.     
    2. RFXSTDTC and RFXENDTC: This pair of variables defines a consistent reference period for all interventional studies and is not open to customization. RFXSTDTC and RFXENDTC always represent the date/time of first and last study exposure. The study reference period often duplicates the reference period defined in RFSTDTC and RFENDTC, but not always. Therefore, this pair of variables is important as they guarantee that a reviewer will always be able to reference the first and last study exposure reference period. RFXSTDTC should be the same as SESTDTC for the first treatment element described in the SE dataset. RFXENDTC may often be the same as the SEENDTC for the last treatment element described in the SE dataset.
    3. RFCSTDTC and RFCENDTC: This pair of variables is used only when the study uses a protocol-specified challenge agent to induce a condition that the investigational treatment is intended to cure, mitigate, treat, or prevent. RFCSTDTC and RFCENDTC always represent the date/time of first and last exposure to the challenge agent.
    4. RFICDTC and RFPENDTC: The definitions of this pair of variables are consistent in every study in which they are used: They represent the entire period of a subject’s involvement in a study, from providing informed consent through the last participation event or activity. There may be times when this period coincides with other reference periods but that is unusual. An example of when these periods might coincide with the study reference period, RFSTDTC to RFENDTC, might be an observational trial where no study intervention is administered. RFICDTC should correspond to the date of the informed consent protocol milestone in Disposition (DS), if that protocol milestone is documented in DS. In the event that there are multiple informed consents, this will be the date of the first. RFPENDTC will be the last date of participation for a subject for data included in a submission. This should be the last date of any record for the subject in the database at the time it is locked for submission. As such, it may not be the last date of participation in the study if the submission includes interim data.
  • No labels