Author: Genpro Statistical Programming Team
Trial Summary Domain [TS domain] – A jigsaw puzzle among the SDTM Trial domains where we need to put together the pieces of the study in a structured way to get a complete picture. This domain is often challenging either in determining the number of parameters required or in getting the required value of the parameters. This blog is a complete guide to the so-called “head-ache TS”. We will discuss the complete structure, required parameters, and the source of values for the parameters.
The TS domain is a trial-level domain that helps to get a complete overview of the clinical trial such as the protocol title, phase, and trial objectives. FDA made TS a “must-have” item for e-submission, even if the study started prior to December 17, 2016. The trial Summary domain has changed significantly since SDTM IG v 3.1.3. New variables have been added to have more control over data. Information presented in a standardized way helps reviewers to have a complete idea of the study quickly and easily.
Challenges
Creation of TS dataset is often troublesome for many statistical programmers.
- We don’t have a one pot solution for all the parameters. Information of TS parameters comes from different sources:
1. Study Protocol (trial phase, protocol title, trial objectives).
2. Clinical Datasets (Study start /end date, Actual Number of Subjects).
3. Dictionaries or terminologies (such as CDISC CT, Registry Identifier, Unique Ingredient Identifier).
- Collecting the above information is manual and time-consuming.
Structure
A trial design domain that contains one record for each trial summary characteristic. This domain is not subject oriented. The domain is composed of fixed structure and a list of variables which cannot be extended. We can see the same in section 7.4.2 of SDTMIG v 3.3.
Like other general observation class, TS domain also have STUDYID and DOMAIN. TSSEQ is slightly different from what it means for other domains in general observation class. TSSEQ indicates just the several possible values of trial summary parameters.
The name, code name and value of the parameters are stored in TSPARM, TSPARMCD and TSVAL respectively. To make the minimum set of variables consistent across all sponsors, new variables have been included from SDTMIG v 3.1.3 onwards. These variables are TSVALCD, TSVALNF (if applicable), TSVCDREF, TSVCDVER.
- TSVALNF stores value called “null flavor” used mainly when TSVAL cannot be filled.
- TSVALCD stores any controlled terminologies associated with parameter’s values.
- TSVCDREF holds the name of the reference terminology from which TSVALCD is taken.
- TSVCDVER contains information about the version of the reference terminology.
Unlike the general observation classes, if the value in TSVAL has more than 200 characters, additional values can be stored in TSVAL1-TSVALn based on the length.
Putting together the Puzzles
By now, we have an overview of how the TS Domain looks like. The one thing that makes TS domain complete is TSPARM. But for this we need an enthusiastic and research-oriented mind.
Populating TSVAL for some TSPARM(CD)’s need consulting the protocol, SAP and other study related documents. Some TSVAL requires controlled terminologies while some can be entered without any controlled terminologies like the study title. Some other requires external sources like clinicaltrials.gov, NCI Term Browser, dnb.com, FDA Substance Registration System etc.
The following table provides detailed information of all the required TSPARM needed as part of submission. We make the process easier to map the TS domain by providing information’s like the expected values, source of the parameter value, any special considerations etc. As mentioned in SDTMIG v 3.3, TSPARM and TSPARMCD are extensible so we can have more parameters added based on the study design.
The column Desired states whether the parameter is required or not. The value “Y” means the parameters must be included and either TSVAL or TSVALNF needs to be populated. The value “Conditional” means the parameter is required based on the study design. For example, if in a study there is protocol specified treatment to be taken in addition to the investigational therapy(ADDON=Y), then the parameter CURTRT, should be included with either TSVAL or TSVALNF populated.
TSPARMCD | TSPARM | TSVCDREF | Desired | TSVAL | Formats/ Codelist | Source or External Link | Additional Notes |
ACTSUB | Actual Number of Subjects | Y | Number | ||||
ADAPT | Adaptive Design | CDISC | Y | No or Yes | Text | Protocol | |
ADDON | Added on to Existing Treatments | CDISC | Y | No or Yes | Text | Protocol | |
AGEMAX | Planned Maximum Age of Subjects | ISO 8601 | Y | Text | Protocol | If maximum age specified is 80 years, then TSVAL=P80Y | |
AGEMIN | Planned Minimum Age of Subjects | ISO 8601 | Y | Text | Protocol | If minimum age specified is 18 years, then TSVAL=P18Y | |
COMPTRT | Comparative Treatment Name | UNII | Y | SRS Preferred Substance Name (or Device Name) | Text | https://fdasis.nlm.nih.gov/srs/srs.jsp | |
CRMDUR | Confirmed Response Minimum Duration | ISO 8601 | Conditional | Text | Protocol | Follow the same format as in AGEMAX | |
CTAUG | CDISC Therapeutic Area User Guide | Conditional | Text | https://www.cdisc.org/ | Include all specific Therapeutic Area Guides used, if any. Use as many rows as needed. | ||
CURTRT | Current Therapy or Treatment | Conditional | SRS Preferred Substance Name (or Device Name) | Text | https://fdasis.nlm.nih.gov/srs/srs.jsp | Use when ADDON=Y. Use as many rows As needed. | |
DCUTDESC | Data Cutoff Description | Y | Text | Describe the basis of data cutoff like “Database lock”. We can use the variable TSGRPID to relate DCUTDTC to DCUTDESC. | |||
DCUTDTC | Data Cutoff Date | ISO 8601 | Y | Text | If data cut-off is based on database lock Then give the date of lock. We can use the variable TSGRPID to relate DCUTDTC to DCUTDESC.. | ||
EXTTIND | Extension Trial Indicator | Y | No or Yes | Text | Protocol | ||
FCNTRY | Planned Country of Investigational Sites | CDISC | Y | ISO 3166-1 alpha-3 Country Code | Text | https://en.wikipedia.org/wiki/List_of_ISO_3166_country_codes | Use as many rows as needed. |
HLTSUBJI | Healthy Subject Indicator | CDISC | Y | No or Yes | Text | Protocol | |
INDIC | Trial Disease/Condition Indication | SNOMED | Conditional | SNOMED CT | Text | https://ncit.nci.nih.gov/ncitbrowser/pages/vocabulary.jsf
| For a healthy volunteer study, TSVALNF = ‘NA’. |
INTMODEL | Intervention Model | CDISC | Conditional | Intervention Model | Text/C99076 | Protocol/ https://clinicaltrials.gov/ | Used when STYPE =” INTERVENTIONAL” |
INTTYPE | Intervention Type | CDISC | Conditional | Intervention Type | Text/C99078 | Protocol/ https://clinicaltrials.gov/ | Used when STYPE=” INTERVENTIONAL” |
LENGTH | Trial Length | ISO 8601 | Y | Text | Protocol | Planned length of observation for a single subject. Follow same format as AGEMAX | |
NARMS | Planned Number of Arms | Y | Number | Protocol | |||
NCOHORT | Number of Groups/Cohorts | Y | Number | Protocol | |||
OBJPRIM | Trial Primary Objective | Y | Text | Protocol | Use as many rows as needed. | ||
OBJSEC | Trial Secondary Objective | Y | Text | Protocol | Use as many rows as needed. | ||
OUTMSEXP | Exploratory Outcome Measure | Conditional | Text | Protocol | Use as many rows as needed. | ||
OUTMSPRI | Primary Outcome Measure | Y | Text | Protocol | Use as many rows as needed. | ||
OUTMSSEC | Secondary Outcome Measure | Conditional | Text | Protocol | Use as many rows as needed. | ||
PCLAS | Pharmacologic Class | Conditional | MED-RT | Text | https://ncit.nci.nih.gov/ncitbrowser/pages/vocabulary.jsf?dictionary=MED-RT&version=2020_09_08 | Use if STYPE = ‘INTERVENTIONAL’ and where applicable for INTTYPE. | |
PDPSTIND | Pediatric Postmarket Study Indicator | Y | No or Yes | Text | Protocol | ||
PDSTIND | Pediatric Study Indicator | Y | No or Yes | Text | Protocol | ||
PIPIND | Pediatric Investigation Plan Indicator | Y | No or Yes | Text | Protocol | ||
PLANSUB | Planned Number of Subjects | Y | Number | Protocol | |||
RANDOM | Trial is Randomized | CDISC | Y | No or Yes | Text | Protocol | |
RANDQT | Randomization Quotient | Conditional | Number | Includes percentage of subjects that are planned to be exposed to investigational treatment. For example, if in a study the subjects are randomized in 1:1:1 ratio and one of the treatment is Placebo then TSVAL is 2/3=0.67. Value ‘1’ denotes all subjects randomized to the Investigational treatment. | |||
RDIND | Rare Disease Indicator | Y | No or Yes | Text | Protocol | ||
REGID | Registry Identifier | ClinicalTrial s.gov | Y | Identifier | Text | ClinicalTrials.gov or EUDRAC | Use as many rows as needed. |
RLPSCRIT | Relapse Criteria | Conditional | Text | Protocol | |||
SDMDUR | Stable Disease Minimum Duration | ISO8601 | Conditional | Text | Protocol | ||
SENDTC | Study End Date | ISO 8601 | Y | Text | The date on which the final data item for a clinical Study was collected from the last study participant (that is,last subject, last visit, or as otherwise defined in the study protocol). | ||
SEXPOP | Sex of Participants | CDISC | Y | Sex of Participants | Text/C66732 | Protocol | |
SPONSOR | Clinical Study Sponsor | DUNS | Y | Data Universal Numbering System or D-U-N-S(DUNS) | Text | https://www.dnb.com/ |
|
SDTMVER | SDTM Version | Y | Text | The value should be the exact term listed in the FDA Data Standards Catalog. If multiple SDTM Versions are used for a study, each version should be listed as separate rows. | |||
SDTIGVER | SDTM IG Version | Y | The value should be the exact term listed in the FDA Data Standards Catalog. If multiple SDTM Versions are used for a study, each version should be listed as separate rows. | ||||
SSTDTC | Study Start Date | ISO 8601 | Y | Text | https://clinicaltrials.gov/ | The earliest date of informed consent among any Subject (Date/Time of Informed Consent, RFICDTC) that enrolled in the study. For studies conducted without informed consent (ie. emergency use) use the date of treatment. Dates for subjects who were screen failures are not included. | |
STYPE | Study Type | CDISC | Y | Study Type | Text/C99077 | https://clinicaltrials.gov/ | |
TBLIND | Trial Blinding Schema | CDISC | Y | Trial Blinding Schema | Text/C66735 | ||
TCNTRL | Control Type | CDISC | Y | Control Type | Text/C66785 | Protocol | |
TDIGRP | Diagnosis Group | SNOMED | Conditional | SNOMED CT | Text | https://ncit.nci.nih.gov/ncitbrowser/pages/vocabulary.jsf | Use when HLTSUBJI = ‘N’ |
TINDTP | Trial Indication Type | CDISC | Conditional | Trial Indication Type | Text/C66736 | Protocol/https://clinicaltrials.gov/ | |
THERAREA | Therapeutic Area | CDISC | Y | Text | Protocol/https://clinicaltrials.gov/ | ||
TITLE | Trial Title | Y | Text | Protocol/https://clinicaltrials.gov/ | |||
TPHASE | Trial Phase Classification | CDISC | Y | Trial Phase Classification | Text/C66737 | ||
TRT | Investigational Therapy or Treatment | UNII | Conditional | SRS Preferred Substance Name (or Device Name)
| Text | https://fdasis.nlm.nih.gov/srs/srs.jsp | Use If STYPE = ‘INTERVENTIONAL’. |
TTYPE | Trial Type | CDISC | Y | Trial Type | Text/C66739 | Protocol/https://clinicaltrials.gov/ |
For QT submissions or important information related to ECG is available, we can include TSPARM related to ECG. The above table describes the required and most used TSPARMs. Also, there are more TSPARMs available in SDTM CT, which we can include based on our study.
The NULL FLAVOR (TSVALNF)
There are situations where we cannot find values for the required parameter TSVAL. If we keep TSVAL column blank, pinnacle 21 will throw an error. What can be done in this situation? This is where TSVALNF comes into picture.
The values that can be used to populate TSVALNF called null flavour, provides additional information when TSVAL is null. The controlled terminology for this null flavour is in ISO 21090 standard, “Health Informatics – Harmonized data types for information exchange. Consider the TSPARAM’s AGEMAX and AGEMIN which are required parameters. The main reason behind inclusion of NULL FLAVOR in this scenario is that most of the protocol do not specify an upper limit for the age of subject. So, we cannot populate TSVAL and hence TSVALNF needs to be filled. There are around fourteen items/values in NULL FLAVOR which are detailed out in section 7.4.2.1 of SDTMIG v3.3. The most used values of NULL FLAVOR are listed below:
- PINF (Positive infinity) – Positive infinity of numbers. Usually used when there is no maximum age specified
- UNK (Unknown) – A proper value is applicable, but not known.
- NAV (Temporarily unavailable) – Information is not available at this time, but it is expected that it will be available later. For example, study end date when the study is ongoing.
- NA (Not applicable) – No proper value is applicable in this context.
Please be careful that both TSVAL and TSVALNF are not populated at the same time. TSVALNF is populated only when corresponding TSVAL is not available.
CONCLUSION
SDTM Trial summary domain provides a high-level overview of the clinical trial. It is an important part of e-submission. Besides being a confusing domain, once created this is the most informative Trial Domain as we get a full picture of the Clinical trial from a single piece. The complete information on TS domain can be found in SDTMIG v3.3 and FDA Study Data Conformance Guide. Appendix B which includes the Required TS parameters for submission will be excluded from SDTMIG v3.4 onwards and will be completely available in FDA Study Data Conformance Guide.
REFERENCES
- Study Data Tabulation Model Implementation Guide v 3.3
- https://www.pharmasug.org/proceedings/2016/DS/PharmaSUG-2016-DS11.pdf