Application of SDTM Trial Design at GSK. 9 th of December 2010

Transcription

1 Application of SDTM Trial Design at GSK Veronica Martin Veronica Martin 9 th of December 2010

2 Contents SDTM Trial Design Model Ti Trial ldesign datasets t Excel Template for Trial Design 2

3 SDTM Trial Design Model Defines a standard structure for representing the kind of high level trial design usually represented in a study schema inaprotocol Arms planned paths through the trial, closely tied to treatment groups or treatment sequences Epochs, Elements new names for periods or phases Visits Provides a dataset for text of all eligibility criteria I/E dataset only includes exceptions some criteria may not have any exceptions Provides a dataset t and terminology for summary descriptors E.g., Title, Phase, Indication, Planned number of subjects Also defines a way to capture actual subject progress that can be compared against the plan For detailed definitions, refer to the SDTM Implementation Guide located on the CDISC website. 3

4 Five Trial Level Datasets The Trial Elements, Arms and Visits describe the planned design of the study. Domain names for these datasets are TE, TA, and TV. Trial Elements Trial Arms Trial Visits The Trial Inclusion/Exclusion c o Domain (TI) is not subject oriented. It contains one record for each of the inclusion and exclusion criteria for the trial. Trial Inclusion / Exclusion Criteria The Trial Summary Domain (TS) is not subject oriented. It contains one record for each trial summary characteristic and is used to record basic information about the trial, e.g., trial phase, protocol title and design objectives. Trial Summary 4

5 Two Subject Level Datasets While the Trial Elements, Arms and Visits describe planned design of the study, it is also necessary to collect actual treatments undergone at visits for each subject. Subject Elements (SE) Subject Visits (SV) SE and SV are based on subject data which already exists. However, preparing these datasets and will require programming decisions E.g., how do we define the beginning of screening? The end of a subject s treatment epoch? What collected data can be used to derive these date/times? 5

6 Excel template for SDTM Trial Design GSK created an Excel template to facilitate the creation of Trial Design datasets The template provides a consistent means of building the five Trial Design datasets (TE, TA, TV, TI, TS). The template includes validations based on the version of the CDISC SDTM Implementation Guide and approved CDISC terminology packages. Template Assumptions: The person entering data into the Excel template understands the protocol and the data that was collected. Visit number assignments reflect a clear understanding of planned visits (those included in the TV dataset) versus unplanned visits. Entering visit start and end rules describe this distinction, and allow the construction of SV, which lists all of a subject s planned and unplanned visits. 6

7 How to use the template TI (Trial Inclusion/Exclusion) is completed by copying eligibility criteria from the protocol. Template holds complete text and a 200-character abbreviation TS (Trial Summary) worksheet includes CDISC terminology, i.e., paramaters and, where appropriate, valid values The Trial Matrix worksheet auto-populates much of TA and TE Easier to visualize trial design in Trial Matrix format TA (Trial Arms) is completed by adding in branching and transition information Branches include randomization, other treatment assignments Transitions include rules for optionally skipping some elements TE (Trial Elements) is completed by adding start and end rules TV (Trial Visits) is populated with visit names and numbers, start and end rules for planned visits. 7

8 User feedback It is pretty good for user-friendliness - when used in conjunction with the Template User Guide far easier than creating these files on an ad hoc basis. It gets easier after the first study There are 4 sets of fields that are common between the TD domains and the ecrf domains, and the issues around getting g them synched with respect to consistent content: TA/DM.ARM(CD), TI/IE.IETEST(CD), SV/TV.VISIT(NUM) and TA.EPOCH versus all other domains. Timings for creating the TD domain vs ecrf data what comes first? Are the TD domains required when running the ecrf data through OPENCDISC checks? Teams require guidance on what to enter on abstract things like if there is no response for a particular item, it s not always clear whether cells should be left empty or entered as none or n/a. 8