Assumptions for Observation Classes

The following assumptions are appliable to domains in the Interventions, Events, and Findings classes. Assumptions in this section will be used with detailed assumptions for appropriate domains.

Assumptions for Interventions Class Domains

Num	Collection Variable Use	Implementation
1	--YN	Variables with the question text "Were there any interventions?" (e.g., “Were there any procedures?”, “Were there any concomitant medications?") are intended to assist in the cleaning of data and in confirming that there are no missing values. Values collected for these fields are not represented in subsequent tabulation datasets.
2	--CAT, --SCAT	Values are generally not entered on the CRF. Implementers may pre-populate and display these category values to help individuals involved in data collection understand what data should be recorded on the CRF. Implementers may also prepopulate hidden variables with the values assigned within their operational database. Categories and subcategories are determined per protocol design and can be populated during tabulation dataset creation.
3	Date and Time	Date variables (e.g., --DAT, --STDAT, --ENDAT) are concatenated with CDASH time variables (e.g., --TIM, --STTIM, --ENTIM, if time is applicable) into the appropriate tabulation --DTC variables (e.g., --DTC, --STDTC, --ENDTC) using ISO 8601 format. Collecting the time an intervention was started is only appropriate if it can be realistically determined and if there is a scientific reason for needing to know this level of detail.
4	-- REASND	--REASND is used with tabulation variable --STAT only. The value "NOT DONE" in --STAT indicates that the subject was not questioned about the intervention or that data were not collected; it does not mean that the subject had no interventions.
5	--SPID	--SPID may be populated by the applicant's data collection system. If collected, it can be beneficial to use an identifier in a data query to communicate clearly to individuals involved in data collection the specific record in question. This field may be populated by the applicant's data collection system.
6	Coding	When free-text intervention treatments are recorded, the location may be included in the --TRT variable to facilitate coding (e.g., lung biopsy). Location may be collected when the applicant needs to identify the specific anatomical location of the intervention. This location information does not need to be removed from the verbatim --TRT when creating tabulation datasets. The nonstandard variables --ATC1 through --ATC5 and --ATC1CD through --ATC5CD are used only when the intervention is coded using the World Health Organization's Anatomical Therapeutic Chemical (ATC) classification system (https://www.who.int/medicines/regulation/): 1 = the anatomical main group, 2 = the therapeutic main group, 3 = the therapeutic/pharmacological subgroup, 4 = chemical/therapeutic/pharmacological subgroup, 5 = chemical substance. The implementer may also add MedDRA coding elements as NSVs to the Interventions domain if this dictionary is used for coding.
7	Location (--LOC) and related variables (--LAT, --DIR, -- PORTOT)	Because the complete lists of controlled terminology for these variables may be too extensive to be relevant for a particular study CRF, sponsors may choose to include only subsets of the controlled terminology on the CRF. --LOC could be a defaulted or hidden field on the CRF for prespecified [--TRT]/Intervention Topic].
8	Relative Timing	Applicants will define the study reference period in the Demographics (DM) domain using TIG tabulation variables RFSTDTC and RFENDTC. Other applicant-specified reference time points can be defined for other data collection situations. The TIG Collection or CDASH Model variables --PRIOR and --ONGO may be collected in lieu of start date or end date. The CDASH variable --PRIOR is used to indicate if the --TRT (the topic item) started prior to either the study reference period or another applicant-defined reference time point. When the study reference period is used as the anchor, --PRIOR may be used to derive a value (from the Controlled Terminology codelist STENRF) into the SDTM relative timing variable --STRF. When populating --STRF, if the value of --PRIOR is "Y", the value of “BEFORE” may be mapped to --STRF. The value in DM.RFSTDTC serves as the anchor for --STRF. When a reference time point is used instead of the study reference period, --PRIOR may be used to derive a value into the SDTM relative timing variable --STRTPT. If the value of --PRIOR is "Y", the value of "BEFORE" may be derived into --STRTPT. --STRTPT must refer to the "time point anchor" as described in --STTPT. The value in --STTPT can be either text (e.g., "VISIT 1") or a date (in ISO 8601 format). The CDASH variable --ONGO is used to indicate if the value in --TRT is continuing beyond the study reference period or beyond another applicant-defined reference time point. When the study reference period is used as the anchor, --ONGO may be used to derive a value into the SDTM relative timing variable --ENRF. If the value of --ONGO = "Y", the value of "AFTER" may be mapped to --ENRF. When a reference time point is used instead of the study reference period, --ONGO may be used to derive a value into the SDTM relative timing variable --ENRTPT. If the value of --ONGO is "Y", the value of "ONGOING" may be mapped to --ENRTPT. --ENRTPT must refer to the “time point anchor” as described in --ENTPT. The value in --ENTPT can be either text (e.g., "TRIAL EXIT") or a date (in ISO 8601 format).

Assumptions for Events Class Domains

Num	Field or Variable	Guidance
1	--YN	Variables with the question text "Were there any <events>?" (e.g., “Were there any adverse events?”, “Were there any healthcare encounters?”) are intended to assist in the cleaning of data and in confirming there are no missing values. These questions can be added to any CRF in order to capture this information.
2	--CAT, --SCAT	Values are generally not entered on the CRF by sites. Implementers may prepopulate and display these category values to help site personnel understand what data should be recorded on the CRF. Implementers may also prepopulate hidden variables with the values assigned within their operational database. Categories and subcategories are typically evident from the protocol design, and could be populated during tabulation dataset creation.
3	Date and Time	CDASH date variables (e.g., --DAT, --STDAT,--ENDAT) are concatenated with CDASH time variables (e.g., --TIM, --STTIM, --ENTIM, if time is collected) into the appropriate SDTM --DTC variables (e.g., --DTC, --STDTC, --ENDTC) using ISO 8601 format. Collecting the time of an event is only appropriate if it can be easily obtained and if there is a scientific reason, such as the need to know the order of events.
4	--COCCUR	--COCCUR may be used when a specific event is solicited (preprinted) on the CRF and the CRF uses an applicant-defined codelist. An applicant may combine the concepts of the CDASH OCCUR variable while also allowing for a "NOT DONE" response. Because the SDTM Controlled Terminology for --OCCUR only includes "N", "Y", and "UNKNOWN" responses, if the CDASH variable --OCCUR is used, the CRF would require a second question to indicate that the data were not collected. The CDASH variable --COCCUR is only used when events are prespecified.
5	--REASND	--REASND is used in conjunction with SDTM variable --STAT. The value "NOT DONE" in --STAT indicates that the subject was not questioned about the event or that data was not collected; it does not mean that the subject had no events.
6	--SPID	--SPID may be populated by the applicant's data collection system. If collected, it can be beneficial to use an identifier in a data query to communicate clearly to the site the specific record in question. This field may be populated by the applicant's data collection system.
7	Coding	The CDASH variables used for coding are not data collection fields that will appear on the CRF itself. Applicants will populate these through the coding process. When free-text event terms are entered, the location may be included in --TERM to facilitate coding and further clarify the event. This location information does not need to be removed from the verbatim term when creating tabulation datasets. The CDASH variables --LLT, --LLTCD, --PTCD, --HLT, --HLTCD, --HLGT, --HLGTCD, --SOC, and --SOCCD are only applicable to events coded in MedDRA.
8	Location (--LOC, --LAT, --DIR, --PORTOT)	Location is collected when the applicant needs to identify the specific anatomical location of the event. Implementers may collect the location information using a subset list of controlled terminology on the CRF. Location variables can be prepopulated as needed. There is currently some overlap across the LOC, LAT, and DIR variables for controlled terminology. While the overlap exists, ensure that this overlap for these variables is not part of database design.

Assumptions for Findings Class Domains

Num	Field or Variable	Guidance
1	--CAT, --SCAT	Values are generally not entered on the CRF by sites. Implementers may prepopulate and display these category values to help site personnel understand what data should be recorded on the CRF. Implementers may also prepopulate hidden variables with the values assigned within their operational database. Categories and subcategories that are not collected and are self-evident from the protocol design are populated during the creation of the tabulation dataset.
2	--PERF, --STAT, --REASND	--PERF defines - variables to record whether an assessment has been performed/collected. --REASND is used to collect a reason why an assessment was not done. --PERF has the Question Text "[Were any/Was the] [--TEST/ topic] [measurement(s)/test(s) /examinations (s)/specimen(s) /sample(s) ] [performed/collected]?" are intended to assist in the cleaning of data and in confirming that there are no missing values. --PERF may be used at the page, panel, or question level. --PERF may be used during the creation of tabulaton datasets to derive a value into the SDTM variable --STAT. The implementer can use a combination of --CAT, --SCAT, with the --TESTCD= "--ALL" and --TEST= "<Name of the CRF module>" to represent what tests were not performed. Applicants must decide how to model each test not performed (e.g., to denote that all tests were not performed using TESTCD = "–ALL"). --STAT has the Question Text "Was the [--TEST] not [completed/answered/done/assessed/evaluated]?; Indicate if (the [--TEST] was) not [answered/assessed/done/evaluated/performed]." This is intended to be used to collect a simple "NOT DONE" check box at the page, panel, or question level. --REASND is used with SDTM variable --STAT only. The value NOT DONE in --STAT indicates that the findings test was not performed.
3	--SPID	--SPID may be populated by the applicant's data collection system. If collected, it can be beneficial to use an identifier in a data query to communicate clearly to the site the specific record in question. This field may be populated by the applicant's data collection system.
4	Date and Time	CDASH variables (--DAT, --TIM) are used in Findings domains to collect the date or date and time that the test was done or performed. The SDTM --DTC variable contains either a date or date and time when a specimen is collected or the start date or start date and time when a specimen is collected over time. Collecting the time is only appropriate if it can be realistically determined and if there is a scientific reason for needing to know this level of detail. Implementers must not use these elements to record a date that is the result of a test. The date of collection of a test may be derived from the date of visit. If so, a separate date of observation field is not required to be present on the CRF.
5	Horizontal (Denormalized) and Vertical Data Structures (Normalized)	In metadata tables, many of the Findings class domains are presented in a normalized structure (1 record for each test) that is similar to the SDTM, even though many data management systems hold the data in a denormalized structure (1 variable for each test). When implementing CDASH in a denormalized structure, create variable names for the Findings --TEST and/or --TESTCD values. To do this, define the denormalized variable names using available CDISC Controlled Terminology for --TESTCD. Alternatively, CDASH variable names for data management systems allowing more than 8-character variable names can use CDASH variables using the following naming convention: <--TESTCD>_<-- SDTM variable name> where --TESTCD is the appropriate CT for the test code (e.g., DIABP_VSORRES, DIABP_VSLOC). In the horizontal (denormalized) setting, SDTM variables such as --PERF, --LOC , --STAT may be collected once for the whole horizontal record and applying the value to all of the observations on that record, or they can be collected per test using the CDASH variable such as <--TESTCD>_--PERF. When tabulation datasets are created, any variables collected for the entire horizontal record must be mapped to each vertical record (as appropriate). In the horizontal (denormalized) setting, an identifier (e.g., --GRPID) may be used to identify all --TESTCD collected on the same record. This facilitates transformation from the CDASH horizontal setting to the SDTM vertical setting and creation of RELRECs.
6	Tests and Original Results	The value in --TEST cannot be longer than 40 characters. The corresponding codelist value for the short test name (at most 8 characters) must also be populated in the SDTM variable --TESTCD. --TESTCD should be used to create a variable name and --TEST be used as the Prompt on the CRF. Both --TESTCD and --TEST are recommended for use in the operational database. --ORRES is used to collect test results or findings in the original units in character format. If the results are modified for coding, the --MODIFY variable contains any modified text. If normal or reference ranges are collected for results, the --ORNRLO and --ORNRHI and --NRIND are used. CDASH does not define the SDTM variable used to standardize the findings results (e.g., --STRESC, --STRESN) or to standardize the normal/reference ranges (--STNRLO,--STNRHI, --STNRC). The standardization of the original findings results and normal/reference ranges is expected to be performed during the creation of tabulation datasets.
7	Location Variables (--LOC, --LAT, --DIR, --PORTOT)	These variables are used to collect the location of the test. The SDTM acknowledges that the results themselves may not be at the same location as the test. This is a known issue with the SDTM. Sponsors may collect the data using a subset list of controlled terminology on the CRF. --LOC could be a defaulted or hidden field on the CRF.
8	–ORRES, --RES, --DESC, and --RESOTH	–ORRES, --RES, --DESC, and --RESOTH collect results for the observation class findings. The variable pairs --RES/--DESC, and --RES/--RESOTH are often used when a finding result is collected using 2 CRF questions but is combined into a single row in the tabulation dataset. --ORRES is used when the finding result is based on a single question and is mapped directly to a single row the tabulation dataset. It is recommended that: --ORRES is used when the finding result is collected using a single question. The result should map directly to the SDTM variable --ORRES. --RES and --DESC are used when a question is asked to collect the finding result, with a follow-up question for a description of the finding. Typically, the question would be “Is the <condition> [normal/abnormal] or [absent/present]?", with a follow-up question such as “What is the abnormality?” or “What is the finding that was observed?” --RES is used to collect whether the finding is normal/abnormal or absent/present and --DESC is used to collect the description of the finding. Typically, this data is modeled in the SDTM as described in the Physical Examination (PE) domain: --ORRES result is normal/absent, --ORRES is the actual abnormality/observed finding and the collected value abnormal/present are not represented. --RES and --RESOTH are used when a question is asked that allows the selection of a pre-specified finding, with a follow-up question to ask about the pre-specified response "OTHER". Typically, the question would be "What is the result?" with a set of prespecified responses, including the choice “OTHER” with the follow-up question “Specify, Other”. --RES is used to collect the finding with pre-specified responses, and --RESOTH is used to collect "Specify, Other". Typically, --RESOTH data is modeled in SDTM as --ORRES (instead of the response "OTHER").
9	Root variables	The Findings About Events and Intervention domains use the same root variables as the Findings domain, with the addition of the --OBJ variable.
10	Adding variables	It is assumed that implementers will add other data variables as needed to meet protocol-specific and other data collection requirements (e.g., as required per protocol, business practice, or operating procedures) to Findings and Findings About domains.

Page tree

Assumptions for Observation Classes