Session 8 Considerations in Creating ADaM Datasets from SDTM Datasets
Session 8 on Thursday, 13 November at the CDISC International Interchange focused on the considerations needed when creating ADaM datasets from SDTM datasets. Topics included pharmacokinetic analyses, the need for human involvement in checking compliance of ADaM and SDTM datasets, best practices for ADaM validation checks, and ADaM Questionnaire (ADQS) datasets. This session was well-attended and sparked many great comments and questions.
The implementation of ADaM in the Pharmacokinetics Department discussed the benefits of standards when analyzing PK data using ADaM datasets. The Statistical Analysis Plan (SAP) should address which data may be excluded from the PK analyses such as concomitant medication use and adverse events which can affect the results. Advantages of standardizing this information in the SAP include increased efficiency, decreased time to create the analyses, as well as increased data transparency (excluded data is clearly highlighted in the SAP, so the analysis results are reproducible).
Another topic discussed was SDTM and ADaM compliance, as well as the needed ADaM validation checks. It is important to check compliance to the standard, but not every convention in the SDTM and ADaM Implementation Guides can be computerized. Those checks that cannot be computerized must be checked by a person who is experienced with the SDTM and ADaM standards.
Common examples of SDTM compliance issues include: mapping data to the appropriate domain; misuse of standard variables such as --CAT and --SCAT; prompt questions in SDTM datasets; analysis data in SDTM (all data needs to be traced back to what data was collected). The Trial Design datasets should also be reviewed against the protocol to insure compliance.
There are over 200 automatable ADaM compliance checks. A key theme of this presentation was “ADaM validation is more than just running a tool and hoping for no errors”. The IG determines the validation checks, not the reverse. Sponsor-defined checks should also be considered when validating ADaM datasets. Language in the IG needs to be explicit if a validation check is to be programmed, vague language cannot be programmed.
ADaM datasets contain many conditionally required variables which are difficult to check programmatically. Traceability of the data back to SDTM is very important. Once findings are identified in the ADaM datasets, any errors that can be fixed, should be fixed. Any finding that cannot be fixed
Strength through Collaboration
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44