Issue Identification and Classification

  • Issues will be identified using the following convention: CAT-### where CAT is a categorical reference defined below and ### is a unique number within category.

Category/Description:

  • MOD = Topics or issues related to the definitions and development of the CDISC models and terminology
  • CDG = Topics or issues related to the CDER Data Guide, previously referred to as the Reviewer’s Guide
  • REV = Topics or issues related to CDER’s review of sponsor submissions
  • IND = Topics or issues related to Industry implementation of CDISC data models

Issues will be prioritized as high, medium, or low and classified as active, resolved, or parked.

Active Issues:

1. MOD-010: High The development of new domains and maintenance of existing domains by the CDISC SDS is not keeping pace with CDER or Industry needs.

  • CDER requirements may result in the agency developing models or constructs that are not aligned with CDISC.
  • Approaches are needed for CDER to move forward in the absence of new standards or stalled standards development.

2. MOD-020: High Adding "analysis-like" content to SDTM skews the intent of the tabulation model and presents traceability challenges.

  • The addition of “analysis-like” variables to SDTM is resulting in a circular conversion process. CDMS data converted to SDTM datasets → Create derived analysis datasets using SAP → derived content from analysis datasets is recycled back into SDTM datasets to make complete.

3. MOD-030: High The SDTM IG does not adequately describe the Exposure domain.

  • EX Assumption 1c provides five different methods for creating/deriving EX.
  • SDTM IG examples are insufficient for industry implementation.

4. MOD-040: High CDISC models do not provide a method for adequately describing the content of supplemental qualifiers or the Findings About (FA) domain.

  • “Forward traceability” – CDISC models should enable the documentation of the content of supplemental qualifiers and FA such that CDER can easily identify key content from those sources that were used in an analysis.
  • If this cannot be done within model, the Data Guide may be a viable alternative.

5. MOD-050: High SDTM does not provide a consistent, scalable solution for relating data.

  • Relating Pharmacokinetic Concentrations (PC) and Pharmacokinetic Parameters (PP) as described in the SDTM IG is often inadequate except for extremely simple PK analyses.
  • Relating microbiology domains in the conversion project has been problematic.
  • --LINKID may not always appropriately relate data points (e.g. censored observations).
  • The relationship FA and their related domains requires RELREC.
  • RELREC, even when understood, is not easily implemented by CDER or Industry. Creating RELREC often requires significant manual processing.
  • CDER does not have tools that use RELREC.

6. REV-060: High CDISC models do not provide a method for distinguishing between findings associated with the same test code.

  • Domains such as LB and VS where multiple findings/units can be associated with the same test code challenge CDER reviewers.
  • Industry (e.g. central laboratories) identifies findings by unique terminology.

7. REV-010: High Supplemental qualifiers cannot be joined to the parent domain.

  • Sponsors are not ensuring referential integrity is maintained between core domains and supplemental qualifiers.

8. REV-020: High Arbitrary assignment of character variable lengths to $200 results in unnecessarily large file sizes.

9. REV-030: High Sponsors often include unnecessary operational content in sponsor-defined domains, core domain variables, supplemental qualifiers, or Findings About.

  • All collected data is not relevant to analysis (e.g. subject initials, laboratory sampling dates).
  • Operational data should not be tabulated and submitted.

10. REV-040: High Arithmetic with ISO 8601 dates and datetimes of varying precision causes CDER’s review tools to fail.

11. REV-050: High SDTM validation errors should be documented in a Data Guide

  • Erroneous validation rules need to be addressed by CDER and Industry

12. CDG-010: High define.XML does not adequately document mapping decisions, sponsor-defined domains, and sponsor-extensions to CDISC controlled terminology in a Data Guide. A standardized Data Guide would help to address this documentation gap.

  • The content and structure of the Data Guide should be standardized and jointly developed between CDER and Industry.

13. IND-010: High Development and maintenance of new CDISC standards will soon result in “legacy” versions of standards data and the potential for a submission of mixed standards.

  • The FDA has stated that submitted data should be traceable to the analysis (i.e. converting legacy data is not required if the analysis was performed using legacy data). Similar guidance for “legacy SDTM” will be needed.

Parked Issues

1. MOD-060: Med SDTM does not provide a standard method for identifying deaths and completers.

  • CDER cannot easily identify deaths and completed in SDTM submission.
  • SDTM IG Amendment addresses identifying deaths and provides precedence for the development of additional variables to identify completers

2. MOD-070: Med Uniquely defining USUBJID across a submission is challenging.

  • The SDTM IG recommendation to define USUBJID as STUDYID-SITEID-SUBJID is being implemented literally; therefore, subjects that are in multiple studies cannot be identified.
  • Rewording the SDTM IG and providing additional guidance for roll-over trials should be considered

3. REV-060: Med Pre-submission planning meetings must occur with the FDA to discuss the submission plan and data standards.

  • The timing, logistics, and content of pre-submission planning meetings is not well defined.
  • Regulatory personnel at sponsor organizations are not the target audiences of the meetings and are not aware that such meetings are necessary.

4. REV-070: Med CDER does not have a common review toolset.

  • The lack of a common review toolset causes issues like differing methods to attach supplemental qualifiers to parent domain.
  • Consider defining a process that enables reviewers, CSS, and industry to provide general use review tools such as SAS macros.

5. REV-080: Med Sponsor’s biometrics staff is often unclear about how CDER performs reviews and the toolsets are available to the reviewers.

  • A lack of understanding leads to incorrect assumptions within the industry. This results in submissions that conform to SDTM, but do not meet CDER's expectations.

6. IND-020: Med SDTM conversions require Data Management-like attention to accuracy and traceability and Statistical Programming analytical mindset and analytical toolset.

  • The Industry is challenged to appropriately staff data conversion activities (e.g. high-cost statistical programmers mapping demographics vs. low-cost clinical programmers deriving treatment emergent flag).
  • CDER may not understand industry processes and lead-time required to prepare data for analysis and submission.

7. IND-030: Med CDER requires Trial Design Model (TDM) datasets.

  • CDER has recognized the potential for the TDM datasets, especially TS, to provide characterizations of studies that CDER can use to identify studies of interest within a data pool or data repository, to address some of the agency’s key mandates.
  • Industry implementation of TDMs is not consistent.