WG5 Subgroup1 F2F March 19-20 2012
When: 19MAR2012, 20MAR2012
Place: Silver Spring, MD
Facilitator: Mark Walderhaug, Mary Nilsson
Scribe: Mary Nilsson
|Beverly Hayes||Janssen Pharmaceutical||1A1B|
|Francois Vandenhende||Clinbay CRO||NA|
|Frank Senk||Astra Zeneca||NA|
|Jack Shostak||Duke Clinical Research Institute||1A1B|
|John Adams||Boehringer Ingelheim||1A|
|Mark Gummel||Theorem Clinical Research||NA|
|Mary Nilsson||Eli Lilly and Company||1B|
|Michelle Barrick||Eli Lilly and Company||1B|
|Samantha Warden||Merck Serono||1A|
|Scott Runyan||Integrated Clinical Systems Inc||1A1B|
|Tineke Callant||SGS Belgium NV - Life Sciences Services||1B|
|Ujjwala Powers||Human Genome Sciences Inc||1B|
|Vincent Amoruccio||Alexion Pharmaceuticals||NA|
Agenda: Subgroup 1 is charged with describing "What Scripts" we want/need in the platform
- Who's here?
Those who signed up for WG5 for the PhUSE conference, and who chose to participate in the Subgroup 1 topic during the break-out sessions of the conference.
The scope for scripts was discussed during the conference. Scope includes Phase 1-4 clinical trial data, able to work with SDTM and/or ADaM data structures, and those scripts that can be considered "common" across therapeutic areas in addition to scripts that may be helpful while reviewing clinical trial data.
The discussion during the 2 days focused on the scope of the scripts. There was general agreement that the focus is on scripts that would apply to Phase 1-4 clinical trial data. However, that's not to say that other areas couldn't benefit from some of the content. There was also general agreement that the scripts should work with SDTM and/or ADaM data structures. It was noted that some scripts could be created to work with "any" data structure, which is certainly fine as well. There was general agreement that most (if not all) scripts would likely create a "static" table, figure, or listing (as opposed to a product that has drill-down capabilities available in products such as Spotfire). We had a lot of discussion about whether we really want a broad set of scripts, or if we should focus on defining a set of scripts that can be considered common across therapeutic areas (eg, define a set of common safety tables, figures, and listings). We decided both are important. FDA reviewers would like to pick from a broad palette of scripts while reviewing data that would include scripts beyond what might normally be done for every product. There is also a need to define the common set to improve efficiency. The FDA and companies could spend less time confirming and validating the common set. Companies can also feel better about knowing they are producing what the FDA is interested in. We decided we should move forward with the "broad set" and the "common set" in parallel. The wording we discussed about this "common set": Consistent approach to common questions (including scripts, visible meta-data, displays).
We are assuming we will have some governance and meta-data.
We want a process for soliciting scripts to answer questions.
Want a way to categorize scripts.
Noted that it is common to have a Summary Dataset between ADaM and the display.
Want test databases.
We want our group to populate scripts first to serve as good examples.
Would we include validation scripts (eg, scripts checking for duplicate data)?
Need clarification on who will actually use these.
Need clarification on purpose and problems we are trying to solve.
We hope the FDA will tell us what they want/need for the "common set". (Industry will likely need to start the "common set" effort and hope FDA will provide feedback.)
Need "end user requirements".
Nice to have a "user interface".
What happens if ADaM changes?
The "common set" can possibly be organized by data domain (categorical, ordinal, continuous, time to event).
Need to continue to work with the ADaM group to ensure we don't duplicate efforts.
Need to start with medical questions we are addressing for the "common set".
Need basic rules on how to make it "usable".
Consider using wikipedia model for "broad set". It's a free for all with monitoring.
There is concern that people may use code without sufficiently thinking.
For the "common set", we know that we won't get this set "required" (which is why we are shying away from the use of the word "standard").
Noted that ADaM and CDASH lack sufficient detail which could cause problems for this effort. Scripts would need to be clear on assumed details.
Mark Walderhaug will meet with those who indicated interest in the "broad set" (1A in the attendance list), and Mary Nilsson will meet with those who indicated interest in the "common set" (1B in the attendance list). However, the co-leads may adjust this plan depending on potential overlap with the other subgroups.
Last revision by MaryNilsson,04/5/2012