Health Level 7
Plenary and Working Group Meeting, September/October 2002
Minutes - Regulated Clinical Research Information Management TC
AGENDA FOR THE WEEK
with annotations from the planning session 9/30 - Q3,Q4
CLICK ON HYPERLINKS TO GO DIRECTLY TO MINUTES FOR SESSION
Monday
30 September 2002
1:45-5:00: RCRIM Strategy and Planning
4:15-5:15 Meet with Wes Rishel
There are a lot of people who want to see HL7 and CDISC harmonized—question is how to do this and are there resources that could be applied. [ODM, SDS, Lab (and ECG)].
Talk through the different models and how to approach harmonizing and dealing with technical implementations other than XML.
Submission stuff doesn’t fit into either message or document—just column name and some attributes—how are metadata dealt with in HL7.
Tuesday
01 October 2002
9:00 am - 5:00 pm
9-10:30: Standard Domains—Publication of SDS Version 2.1 Domains as an HL7 Document
focus on elements rather than technical implementation
9:00-9:30: Peter Covitz and Denise Warzel of NCI will give a presentation on their current efforts with their Metathesaurus and data interchange efforts. Objective - find right interface between NCI efforts and HL7
Current thinking of FDA—regroup SDS domains into three categories: findings, interventions, and events. May be complimentary to NCI data element work.
11-12:30: Standard Domains—ECG
Barry Brown will present the current version of the ECG model. Barry presented a proposed XML message for plot last time. Since then he has been working on recasting it as an HL7 message, looking at ways the RIM may need to be adjusted and how it could potentially be extended and used for any kind of graphical data. Barry mentioned that he needs to work with someone to incorporate everything that needs to there for clinical trial.
1:45-3:00: Standard Domains—Lab/ODM
Labs update and overview of RIM/HL7 harmonization efforts.
3:30-5:00: Standard Domains—Joint with Protocol Representation Sub-team
Update and overview of progress by protocol representation team. Want to ensure that protocol representation team dovetails with what has gone on with SDS study standards. Norman Stockbridge will attend and perhaps can discuss some of the issues he observed at DIA Data Standards meeting.
Wednesday
02 October 2002
9:00 am - 5:00 pm
9-10:30: - Protocol Representation working session
- Clinical Genomics SIG
The new Clinical Genomics SIG will meet from 0900-1230. This SIG is parented by Orders & Observation, but is of interest to RCRIM as well. Susan Bassion and Barbara Tardiff are planning to attend. Barbara will present an overview of RCRIM.
02 October 2002 (continued)
11-12:30: - Protocol Representation working session continues
- Post-market AE Reporting
Mead Walker has mapped E2B specifications to V. 3 RIM; FDA looking at this as electronic Medwatch. Will present and discuss these efforts.
1:45-3:00: Protocol Representation working session continues
Post-market AE Reporting continues
3:30-5:00: Post-market AE Reporting continues
Thursday
03 October 2002
9:00 am - 5:00 pm
9-10:30: Product Labeling
Sandy Boyer will present work on what she has been doing with product labeling (under contract with FDA). FDA wants to distribute label data in machine processable format that could be used to generate stand-alone label. If there was a change in an existing label, manufacturer would only send sections that changed; FDA would disseminate/publish as new version of label.
Bill Hess (FDA) will be attending Vocabulary sessions all week and present what they are working on related to the labeling effort.
11-12:30: Product Labeling continues
1:45-3:00: Joint with Structured Documents
Discuss Product Labeling proposal
Structured Document template for CDISC Operational Data Model
Use of structured document for stability data (FDA Veterinary medicine has contracted with Gunther to develop message for stability data)
3:30-5:00: Wrap-up—RCRIM Actions items; Review of proposed standards and plans for balloting
Health Level 7
Plenary and Working Group Meeting, September/October 2002
Minutes - Regulated Clinical Research Information Management TC
ATTENDEES
NAME / REPRESENTING / EMAIL / MON / TUE / WED / THU9/30 / 10/1 / 10/2 / 10/3
Bassion, Susan / CDISC / / X
Boyer, Sandy / self / / X
Broverman, Carol / Fast Track System / / X / X / X
Brown, Barry / Montara Instrument / / X / X / X / X
Budnitz, Dan / CDC / / X / X / X
Cassidy, Mike / Siemens / / X
Covitz, Peter / NCI / / X / X
DelSignore, Tom / Trimeris / / X / X / X
Eskandanian, Kolaleh / Georgetown Univ. / / X
Fetvedt, John / IBM / / X / X
Fry, Jim / Infosystems Management / / X / X
Gattadahalli, Satish / VHA/EDS / / X
Goetsch, Roger / FDA/CDER/ODS / / X
Hartel, Frank / NCI / / X / X
Hollinger, Kathy / FDA / / X / X / X / X
Kordik, Steve / GE Medical / / X
Kubick, Wayne / CDISC / / X / X / X
Kush, Becky / CDISC / / X / X / X
Levin, Randy / FDA / / X / X / X / X
Luccioli, Stefano / FDA/CFSAN / / X / X
Newby, Frank / CDISC / X / X
Palmer, Michael / CDISC / / X / X / X
Park, Jayoung / FDA/CFSAN/OFAS / / X
Pollock, Dan / CDC Atlanta / / X
Quade, Linda / Lilly / / X / X / X / X
Stockbridge, Norman / FDA / / X
Tardiff, Barbara / CDISC, Regeneron / / X / X / X / X
Thurmond, Scott / FDA / X
Walker, Mead / Mead Walker Consulting / / X
Wilson, Steve / X
Health Level 7
Plenary and Working Group Meeting, September/October 2002
Minutes - Regulated Clinical Research Information Management TC
MINUTES
30 September 2002: Q3,Q4
Strategy and Planning for week’s meetings:
Reviewed agenda for the week, and objectives for each session.
See Agenda at the beginning of this document for summary of objectives for each session
Discussion with Wes Rishel:
Wayne summarized the three major issue:
1) We have three CDISC models—two don’t fit HL7 message/document model well.
2) Lab it has been fit into HL7 and can be made to conform to HMD.
3) We would like to get HL7 “approval” for multiple technical implementations (XML, SAS, and ASCII) particularly for Lab.
Wes discussed some his thoughts.
One of strengths of XML is that it is hierarchical and extensible. Other technical implementation may have limitations if they do not have these features. HMD is universal description. Ideal would be to write set of rules for each technology that would translate HMD into that implementation technology specifications; advantage is that once something is specified in an HMD, it can be represented in any implementation technology.
A strength of HL7 is that it is accredited by ANSI but that doesn’t come for free. It requires a laborious process to get something accepted as standards. However HL7 does have official documents that are not accredited standards but are recommendations. The question is really how official do we want it to be—with motivated group can go through most of the process electronically—doesn’t take a set number of meetings. In terms of what can get written into a specification—there are no rules, especially when there is a regulatory flavor.
All kinds of methodologies are applied to create the HMD. If there is a business need to create something that doesn’t follow existing methodologies (and have made a good faith attempt to make it fit) then it is not a problem to use another methodology. The XML ITS is an approved specification (set of rules) that can be applied to an HMD to produce a Schema (using plug-in that tells how to go from HMD to Schema). We could develop a SAS ITS but this might take considerable work to make it work for all HL7 HMDs. If time is critical may make sense to come up with a SAS specification that is just limited to the RCRIM message(s).
Area of acceptability/readability of HMD is something HL7 has worked hard at and according to feedback they have made progress. For years they have had this discussion—who cares the machine just has to read it but the reality is that people have to read it to get comfortable with. Wes suggested that after Linda has completed HMD she will send to him and spend half an hour on phone walking him through it. HL7 is working on getting rid of artifacts. It is not clear that HL7 will ever get a general description that will make sense in the specific. The organization is more inclined to try to get translations than to have a whole collection of standards. The benefit of aggregation is that creates universality of expression; meet needs for local and general expression.
SDS is really metadata model and HL7 does not have already established mechanisms of handling metadata. There is a lot of prior art out there as to how to describe metadata and we should look at capitalizing on prior art. Wes suggested we should look at WSDL (web services standard way of defining metadata). If we use WSDL then all sorts of tools and vendors should be able to read and use metadata. WSDL is an XML specification. In Randy’s mind SDS is an information model that represents observations that would be collected as part of clinical study.
Wes indicated that there two relevant issues:
Aggregating data collected at multiple times under different metamodels
Communicating the metamodel
XML and HL7 both good at dealing with the first issue but haven’t yet taken up the second issue.
Based on all this, Standard Domains team will probably go forward with casting the CDISC SDS models as an HL7 Informative document.
01 October 2002: Q1
Standard Domains
Peter Covitz of the National Cancer Institute (NCI) presented an overview of the NCI bioinformatics activities:
caCORE describes the core infrastructure for biomedical application development and includes Thesauri and Ontologies, UML-modeled middleware and APIs, and data standards repository (metadata)
The cancer Data Standards Repository (caDSR) is a metadata management system for NCI common data elements (CDE). This allows those conducting clinical trials to form their own CRFs in terms of layout, however, they must select the CDEs from the caDSR. Curators help maintain the caDSR and harmonize across agencies.
caDSR is implemented in Oracle DBMS. They are building tools to make all of this accessible including tools for “form compliance review”, “browse/query/export”, and a specialized high-end “curation” tool. Programming interfaces include PL/SQL and XML; will soon be able to do export as XML which can be read directly.
The model used to develop CDE is ISO/IEC 11179. This is a generic metadata standards model. It supports both standard and ad hoc groupings (classifications for breast, lung, kidney CA for example); administrative components allow to manage process from curation standpoint (proposed/approved, etc.).
Lilly has done something similar but have not yet automated it to the same extent. CDISC uses spreadsheets (Excel) and FDA uses Word to manage the data elements.
Multiple groups are “making” data elements: CTEP, Biomedical Imaging Program, Cancer Prevention, Cancer Biomarkers Research Group—putting them all into the same system with hope that they will be able to harmonize moving forward.
Currently deploying in Oracle Clinical—but not specific to any one type of deployment. Additional domains to be developed include pediatrics, genomics, and Phase 1 and 2 trials.
NCI is interested in finding more about HL7, how to harmonize with HL7 and what does it mean to have a metadata standard in HL7. Everyone at NCI would like to see this opened up to the larger world and get other folks involved in harmonizing. Right now web interface requires secure log-in but just have to request user name and password from Peter Covitz (). Eventually read-only access will be generally available. NCI is also willing to make tools available to industry at no cost. Peter’s supervisor is Ken Buteow and he will be at Clinical Genomics SIG.
01 October 2002: Q1 (continued)
Wayne Kubick reviewed agenda:
NCI Presentation
Standard Domain Working Group Goals
Introduction to FDA Information Model
Discussion: Moving forward on HL7 clinical research Standard Domains—CDISC, FIM or both
Wayne presented an update of the CDISC SDS work and the FDA Patient Profile Viewer Pilot, with respect to an emerging Information Model approach to modeling the SDS data. The SDS_v3 model combines the CDISC SDS approach and content with the HL7 concepts. This model will provide the flexibility needed to characterize the science in clinical studies and permit the seamless transfer of information across systems. Adoption of alternative approaches to representing the safety and efficacy domains is described in the Scope Statement of the RCRIM Standard Domains group and will facilitate the adoption of these standards as ANSI accredited standards through HL7.
Data structure created for patient profile included subject characteristics, observations, and events.
FDA reworked these structure to roll domains into three categories: Interventions, events, findings. This new structure was referred to in this meeting as FIM (FDA Information Model), but this is not the preferred name going forward. Each finding in FIM has topic variable, subject id variable, timing variable, and qualifier (severity) variable. Fit timings variables based on what makes sense. This is being considered as the design for an FDA data repository as well.