Bay Area CDISC Implementation Network Meeting – Thursday, November 20, 2008
Q & A Meeting Minutes:
ADaM and ISS
Revathy Duvedi of ICON discussed “Using ADaM Datasets for an ISS Submission.”
(1)Q: Why did you choose an ADaM structure for the analysis datasets rather than an SDTM+ structure?
A: ADaM is one proc away. I prefer an ADaM structure because there are many small datasets compared to large, derived datasets.
(2)Q: What is the ADAE max flag?
A: It flags the record with the maximum severity if more than one record exists for the same AE.
(3)Q: Did you create ADaM datasets for individual studies?
A: There were 8 studies in the ISS. Individual studies had SDTM created but not ADaM.
(4)Q: Did you retain subject baseline characteristics in ADSLs?
A: Yes, for weight, height, etc.
(5)Q: Have you received any feedback from the submission?
A: No, not yet.
(6)Q: How did you trace back from records in ADaM to records in SDTM?
A: Unique Subject ID has the study number and patient number and I used that information to trace back to SDTM.
(7)Q: How do you handle the unique subject ID if the subject is in more than one study?
A: It is up to the sponsor how to handle it.
(8)Q: Did you encounter any challenges with using ADaM for the ISS?
A: Yes, especially with AE in cross-over studies. How you assign AEs to which dose and period, and the treatment-emergent flag can be challenging. How AEs are collected is key.
(9)Q: Did you have any challenges specifically with using the ADaM to do the ISS?
A: The SAP needs to be thought through and developed with ADaM and SDTM in mind.
(10)Q: Did you use define.xml? Did you have to define variables study by study?
A: Yes, 8 studies were included in the define.xml
(11)Q: Did you use any automatic tools for the metadata?
A: Yes, but I cannot provide the details because the tools were developed by someone else.
CDISC/FDA Integrated Safety Database Pilot Study
Ian Fleming of Genentech talked about “The Second CDISC FDA Pilot Study” and lessons learned from it.
(1)Q: Who is the primary customer?
A: FDA. The FDA will probably compile a safety profile to determine if compounds are safe individually and across classes.
(2)Q: What tools did you use for the pilot study?
A: WebSDM and Octagon Viewpoint. The FDA will be using WebSDM. I am unsure if the FDA will load the pilot into Janus. I am also uncertain if PROC CDISC is being used.
(3)Q: Can you suggest any good strategies for better Metadata Management?
A: Metadata is more complicated and it is better to define it after CDASH is established. Value level metadata, controlled terminology, xml…
(4)Q: Is SAS SDD good to use?
A: I cannot recommend any specific software or vendor.
(5)Q: What is SAS SDD?
A: It stands for SAS Drug Development and it is a metadata management and data repository product sold by SAS.
CDISC International Interchange – Panel Discussion (Members: Mario (Genentech), Erika (Fibrogen), Scott (Genentech), Anne (Amgen))
(1)Q: Any impression or recommendation for attending this conference?
A: This year was the first time attending the interchange for most of the panel members and they all recommended it. They were all in agreement that a lot of information was provided by vendors and presenters but that the content was general and theoretical. There was not a lot presented that could be brought back and immediately implemented. One panel member thought there was not a lot of showing how things were done during the presentations perhaps because they were only 20 minutes long. All panel members concurred more examples would be helpful. One of the panel members who attended the interchange for the second time mentioned the first year all of the information seemed overwhelming but when she returned this year she knew what to expect. She also felt it was a great time to ask people who have been implementing the standards your questions. The panel members felt it was good to have interaction with the FDA and have them involved in the discussion.
(2)Q: FDA Session – any take away messages?
A: (1) The FDA has a vested interest in CDISC. Amy Mala recently joined CBER from industry and she has been very involved in team building and CDISC at CBER. (2) Two models were rolled out this year – the Clinical Data Acquisition Standards Harmonization (CDASH Version 1) for case report form field standards and the Protocol Representation Standard Version 1.
(3)Q: Any statements of policy by the FDA?
A: In general the panel thought the FDA is still trying to get organized with CDISC standards and implementation. One panel member said 2014 is the FDA’s target for CDISC implementation and the HL7 messages.
(4)Q: Exhibitors – any interesting stuff?
A: A presentation was given by a vendor about how to implement ADaM using their software.
(5)Q: Did you get the impression that people were at the early or late stage of implementation?
A: People were at both extremes in terms of their knowledge and experience with implementing the standards. There was more discussion about CDISC initiatives other than SDTM and ADaM this year like CDASH, ODM, etc.
(6)Q: Any other observations to report on?
A: The conference was a good opportunity to learn how the different standards teams fit together and where/when tools will get used. People are still figuring out the standards and how to best implement them.
(7)Q: What kinds of people attend the interchange?
A: Mainly people from DM and Statistical Programming, and occasionally people from regulatory. Some large companies are establishing governance committees for the implementation of CDISC. There were several folks from the FDA and NIH in the audience, not presenters. The Delaware CDISC Network meetings are stuffed with FDA people and they speak openly about their views.
CDASH
Dorothy Dorotheo of InterMune and Alec Vardy of Exelixis gave an update on the CDASH Project and Its Purpose and Scope.
(1)Q: Which domains gave you the most trouble in terms of collecting them so they could be put in a vertical structure?
A: Labs, ECGS, and Vitals because all or portions of the data collected are not always done. Labs caused trouble because the data can come from both Central and Local Labs.
(2)Q: Are there any automatic tools you can recommend for the process?
A: No. However, CDASH was established to make the process easier and later on a more automatic process can be created.
(3)Q: How are NOT DONE, UNK, and NA fields handled?
A: CDASH has some data collection fields not in the SDTM. Some code lists include UNKNOWN, NOT REPORTED, NA. Most situations have been covered in code lists or best practices.
(4)Q: Should we only collect the necessary data?
A: It is related to cost, but a general comment is to make sure the embedded data are consistent. Also, it’s important to work with your statistical group to know how the data will be analyzed.
(5)Q: What domains are highest priority?
A: At the moment only the device domain. The highest priority aside from devices is to focus on early implementers and their feedback.
(6)Q: Is the CDISC protocol team currently working with the CDASH team for CDASH model development?
A: No.