Minutes of Meetings of the Clinical Trials Special Interest Group

HL7 Working Group Meetings – January 2002, San Diego, CA

Meeting Attendees:

TUE / WED / WED / THU
NAME / COMPANY / eMAIL / EVE / Q1/Q2 / Q3/Q4 / Q1/Q2
Linda Quade / Eli Lilly & Co. / / X / X / X / X
Wayne Kubick / CDISC / / X / X / X
Scott Getzin / Eli Lilly and Company / / X / X / X
Barbara Tardiff / CDISC/SAS / / X / X / X / X
Randy Levin / FDA / / X / X / X / X
Rebecca Kush / CDISC / / X / X / X / X
Denise Warzel / National Cancer Institute / / X / X / X
Susan Bassion / CDISC / / X / X
Chuck Jaffe / Astra Zeneca / / X / X
John Fetvedt / IBM / / X / X / X
Julie Evans / CDISC / / X / X / X / X
Lori Baca / Amgen / / X / X
Geoff Gordon / Lincoln Technologies / / X / X / X
Steve Brown / Veterans Administration / / X
Harry Solomon / General Electric / / X
Bob Dolin / X
Norman Daoust / Daoust Associates / / X

Tuesday, 8 January

5:15-6:15 p.m.

Introductory meeting with Decision Support Technical Committee, which currently encompasses four SIGs:

  • Arden Syntax SIG
  • Health Records SIG
  • Clinical Guidelines SIG
  • Clinical Trials SIG (CDISC-HL7 collaboration)

ACTION ITEM: CTSIG should assign a team member to serve on the Decision Support Technical Committee.

6:15 – 9:00 p.m.

Presentation by Linda Quade of her work on representing the CDISC LAB content model within the HL7 Version 3 Reference Information Model (RIM); she has produced much of an rMIM at this point. This is a proof-of-concept effort, to determine the suitability of the representation of the Labs content model through the HL7 methodology, and the CDISC Labs Team’s comfort level with that representation. Following acquisition of the input from the LAB team and possibly others within HL7, the next step would be an HMD. The group was very encouraged to see this progress: all of the content required by the LAB model was accommodated by the HL7 RIM, and the resulting model appeared more familiar to the CDISC group than did the RIM itself. Outstanding issues include reviewing the hierarchy used (e.g. for this version of the model, the observation was the ‘center of the universe’ whereas the LAB team may prefer this to be the clinical trial) and eliminating or masking certain RIM attributes (e.g. those that would provide confidential information on the subject/patient).

ACTION ITEM: The course of the discussion suggested that CDISC could profitably augment their Glossary with terms such as subject and patient and other terms from the CDISC teams. Such definitions are necessary guidance to creating interchange formats that can be used consistently across companies.

Wednesday, 9 January

9:00 a.m. – 12:30 p.m.

Election of Co-Chairs - Randy Levin, Linda Quade, and Barbara Tardiff were elected co-chairs of the CT SIG

Discussion on promoting the SIG to a TC –

Randy Levin of the FDA presented his view of how this committee could support regulatory activities, both in the US and internationally. The breadth of support suggested the need for a full-fledged TC.

The committee worked on a white paper outlining the rationale, scope, projects, etc. for such a committee.

Focus areas could include:

Protocol and Study Documents
  • Regulatory Applications and Standard Submission
  • Standardized Domains for Clinical Research Data
  • Medication Information
  • Surveillance Information

A draft of the white paper is available in the CT SIG documents area of the HL7 website; title: Proposal to Create New HL7 Technical Committee (Jan02).

1:45 – 3:00 p.m.

Presentationby Steve Brown of VA in Nashville, TN regarding Medical Terminology work currently being done with FDA on the National Drug Formulary Reference Terminology (NDF RT) Model. A model is used to predefine attributes and aggregate under headers to form an appropriate hierarchy. Input has been provided from NDA, NLM, FDA, NCI, and Apelon to this NDF terminology. Progress has included tool development, model creation, reference taxonomies, and concept definition.

The NDF RT Model includes:

Chemical Structure Class, Active Ingredients, Clinical Drug, Finished Dosage Form, Product, Packaged Product, Kit;

Structural ID, Strengths/Dose Forms/Routes of Administration, Appearance, Brand Names, National Drug Codes-NDC;

Mechanism of Action, Therapeutic Use, Pharmacokinetics, Identity Key and Database Links.

This information all comes from the labels. (Currently, labeling is outside of the scope of the ICH.) FDA will ensure that this is maintained.

Presentation by Geoff Gordon on ECG standards. A meeting was held at FDA on 19 November 2001 to raise issues and receive comments regarding ECG standards. One objective is believed to be that the FDA needed to raise the awareness that they will be requiring the biopharmaceutical companies to submit electronic wave forms from ECGs with their New Drug Applications and also to archive these in an electronic form (e.g., Holter monitor recordings). Scott Getzin (LAB team) and Geoff Gordon (ODM team) represented CDISC at this meeting. An example of an XML DTD was posted/proposed by FDA for comment. This DTD contains as its elements the information that the FDA would like to have submitted as ECG information, including the actual waveform and relevant information about it.

Scott and Geoff have worked with Wayne Kubick to develop a document of suggested changes to the proposed FDA Timeseries DTD. These changes would enable compatibility with SDM.

Scott further raised the issue of compatibility with existing waveform interchange standards such as those developed by HL7 and DICOM. The committee endorsed by voice affirmation an action item directing Scott, Geoff and Wayne to provide a response recommending the FDA model be based upon existing standards (including the HL7 model, with DICOM modifications and CDISC harmonization with the SDS). This recommendation should be submitted on behalf of CDISC and the CT-SIG. It should be posted on the docket and on the CDISC website. An actual proposed model should then be submitted as soon as possible following the proposed recommendation. Initial paragraph response should be posted to FDA Docket by 21 January, stating that more details will come; by 31 January, there will be a more detailed response.

3:30 – 5:00 p.m.

Presentation by Norman Daoust and Geoff Gordon regarding their project to compare and contrast the CDISC ODM and the HL7 RIM.

Contrasting

  • single-center vs. multi-center trials
  • outcome of NDA vs. journal publication
  • observation vs. randomized trials
  • may follow GCP vs. must follow GCP

Similarities

  • regulated research
  • may wish to have both outcomes of NDA and journal publication
  • data may become ‘re-purposed’ to serve needs of both types of trials
  • desire to use same processes

The Plan – ODM – Existing Data Representation in Sponsored Clinical Trials

  • Written protocol document, CRFs, lab data specified, data dictionary or empty dataset files, verification and analysis plan for the data
  • Operational database based upon the data dictionary (CRF and lab report data in this database, which is the source for the CRTs)
  • Submit to FDA SAS datasets containing CRTs along with PDF files of CRFs and analysis datasets; audit trails available; source data available at sites
  • Concept – single electronic trial document

First Steps in HL7-CDISC ‘Harmonization’

  • Created model to define a flow of how data in a multi-center trial might occur, in an manner that can be archived or ‘snapshots’ of data such as for an interim report
  • Define message vs. document
  • Messages include data, context and instructions; have defined senders and receivers; may include a ‘family of messages’ and an acknowledgement of the receipt of the data
  • Documents include data and context; are intended for sharing information and/or archiving; have no mechanism for real-time transfer or acknowledgements; has creators and users vs. senders and recipients; and could actually allow for further sending without knowing this occurred
  • Define fixed data vs. variable data definitions – ODM does not define standard names for data elements to be collected; HL7 requires use of standard names based upon its model, i.e. RIM attributes or vocabulary codes
  • Recommendations
  • Explore further to define the role and usage of documents vs. messages
  • Define a standard ‘mapping’ for commonly used patient care or clinical trials messages in HL7 and ODM item groups as defined by SDS (e.g. Demog, PE, AE)

Possibilities

  • Merge the two standards
  • Keep the standards separate but define when they should be used

Ultimate Goals (examples)

  • Streamline data exchange among stakeholders
  • Facilitate collection of high quality data for clinical trials at the investigative sites

Thursday, 10 January

9:00 – 10:30 a.m.

Joint meeting with Structured Documents representatives, Robert Dolin, M.D. and Rhonda. Randy Levin presented the FDA/ICH concept of structured documents (i.e. NDAs and the CTD). Bob Dolin responded by indicating how this could fit within the work efforts of the Structured Document Committee and showed a model of their work. Timing was discussed. Randy Levin will need to make a proposal to ICH in this regard since he would like to use the HL7 standard instead of creating another one. A gap analysis between the HL7 Structured Documents header and the FDA header needs to be done to determine how the FDA/ICH requirements could be met by the HL7 standard.

ACTION ITEM: Bob – Send the Medical Records R-MIM to Linda Quade

ACTION ITEM: Linda contact Bob Hizer about the gap analysis between Randy’s header and the Structured Documents header

ACTION ITEM: Sandy Boyer – has volunteered to help with this effort from the Structured Documents group (through Bob Dolin)

11:00 a.m. – 12:30 p.m.

Follow-up on timing of ECG response (i.e. by 21 and 31 January 2002) – see section under 9 January.

Follow-up on ODM-RIM discussions from Wednesday –

Ideas were presented by Geoff Gordon. Strong sense that we need to pursue path to merge standards efforts. Small group should do the next steps.

  • Specification of goals of merged standards
  • Transitioning plan for ODM-based data
  • Actual modeling work

Denise Warzel has a strong interest, especially from NCI side. Barbara Tardiff interested, also, from the perspective of academia and biopharmaceutical development.

Goals for the merged model:

  • Ensure common terminology
  • Enable ‘auto-harvesting’ of some trial data from hospital information systems
  • Enable standardized stat collection at sites built into standard patient record and management software
  • Support ‘data management messaging’, i.e. data verification/clarification/quality
  • Support original ODM requirements, i.e. archiving requirements, ability to transfer data from a whole trial in a single tree of documents (not unlike CDA concept); need to talk with other organizations that participate with HL7 re: representation of entire patient record, e.g. TEPR

Approach to achieve goal of reaching a common standard for clinical trials:

  • Develop an ‘HL7 way’ to define study metadata, including CRFs, data elements and standard structure for viewing export
  • Probably based on ISO-11179 standard
  • May involve adopting ODM metadata model
  • Use this for submissions metadata as well
  • Define clinical and administrative data using extensions of existing HL7 (observation) messages/documents (some aggregation of results)
  • Define framework/extensions to reference metadata
  • Include mechanism for audit train/signatures
  • Define backward compatibility/conversion to/from existing ODM data organization
  • Ultimately, make this an ANSI-accredited/ISO-accredited standard

Action Step: Becky/Shirley to schedule a conference call for initial follow-up on this project among Denise, Linda, Norman, Geoff, Barbara, Julie (Wayne?)

1:45 – 5:00 p.m.

Linda Quade, Scott Getzin and John Fetvedt met with Orders and Observations TC representatives to follow up on mechanisms behind representing the CDISC LAB model in the HL7 RIM. See OO minutes for details of this session.

Randy Levin, Wayne Kubick, Barbara Tardiff and Becky Kush met together in an unscheduled meeting to refine the proposal for the new Technical Committee to be presented in the evening to Wes Rishel and Stan Huff. See discussion above on promoting the SIG to a TC, and referenced document Proposal to Create New HL7 Technical Committee (Jan02) in the CT SIG area of the HL7 website.

7:00 - 9:00 p.m.

Discussions with new HL7 Board Chair (Wes Rishel) and immediate past Chair (Stan Huff) regarding new Technical Committee proposal (above)

Initial concerns voiced (at Board earlier in the week) were potential overlap with existing groups within HL7 and facilitation resources. Looking over the list of projects in the proposal of 10 January, it appears that there is little overlap with most of the projects listed, although there are a number of opportunities to collaborate with other HL7 groups. The area of greatest overlap (and one where there is already ongoing collaboration) would be the Medical Information Terminology. The facilitation issue was discussed with a potential to identify CT-SIG members or others whocould be trained as facilitators. Issues voiced by the CT-SIG representatives at this meeting were the need for support from HL7 to pull in participation from additional groups (e.g. ICH) and the need to ensure sufficient meeting space if we attract additional members for this TC and its task groups.

The outcome of the discussion was:

  • It should be proposed that the CT-SIG be ‘upgraded’ to form this Technical Committee. Linda Quade will fill out the form to be presented to the Technical Steering Committee and steps should be taken per HL7 guidelines in April.
  • Stan Huff and/or Wes Rishel will be in Washington, DC in January and will try to coordinate with Randy Levin a meeting with Janet Woodcock and other FDA representatives to discuss benefits of additional collaboration.
  • Space requirements will be outlined and sent to Wes Rishel by Barbara Tardiff. He will work to try to accommodate the new Technical Committee at the spring meeting in Atlanta and to ensure that HL7 staff understands this is for HL7 (vs. outside groups).
  • Designated responsible parties for the project groups will attempt to pull together additional resources to come to the Atlanta meetings and to further identify the goals of these groups.

HL7 CT-SIG Meetings – Draft Notes1January 2002

R.D. Kush