RCRIM

1
RCRIM Standard Domains Agenda

• NCI Presentation
• Standard Domains Working Group Goals
• Introduction to FDA Information Model (FIM)
• Discussion Moving forward on HL7 CR Standard
  Domains
• CDISC, FIM or both

2
Standard Domains for Clinical Research Data -
Goals

• To identify alternatives for incorporating
  CDISC data models that represent typical safety
  and efficacy data domains used in clinical
  research into HL7 as accredited standards,
  implementation guides, documents, etc. and to
  expand these models with standardized codes and
  controlled terminology
• To develop standard messages for transmitting
  clinical research data (such as lab data)
• Overlaps with CDISC ODM, LAB and SDS teams.

3
Standard Domains for CR Data Discussion Points

• Identifying ways to get CDISC submission domain
  models published as an HL7 document or standard
• Exploring opportunities to utilize HL7 work in
  vocabularies and codes to improve the SDS models
• Consider applying the ISO 11187 standard for
  metadata to the models 
• Promote interaction with multiple groups (CDISC,
  NCI, FDA, CDC-NACCR, Professional Societies)
• Continue current activities to explore an HL7
  implementation of the LAB model and an HL7
  rendition of the ODM.

4
SDS V. 2.0 Domain Models

• Demographics
• Disposition
• Exposure
• Labs (Chemistry, Hematology, Urinalysis)
• Physical Exam
• Medical History

• Concomitant Medications
• Adverse Events
• Vitals (Horizontal)
• Vitals (Vertical)
• ECG (Horizontal)
• ECG (Vertical)

5
Patient Profile Spec vs. SDS 2.0 Key
Differences

• Removes common selection variables
• Can be joined into views by data warehouse
• Uses more normalized Demographics
• Standardizes fixed set of core required variables
• Can represent any additional characteristics in
  separate Subject Characteristics table
• Introduces new Study Summary table
• Will simplify document creation and record key
  attributes of studies (not required for pilot)
• Classification of variable roles Subject
  identifiers, Topic, timing, qualifiers
• Introduces concepts of controlled terminology and
  standardized content requirements

6
Patient Profile Structures
Subject Events
Observations Assessments Attributes/Trtment

Studies
Labs
Exposure
Adverse Events
Vital Signs
Demo- graphics
Subject Characteristics
Findings
ECG
Con Meds
Physical Exam
Disposition
Medical History
7
FDA Information Model Structures
Interventions
Events
Findings
Labs
Studies
Adverse Events
Exposure
Vital Signs
Con Meds
Demo- graphics
Subject Charstics
Medical History
ECG
Disposition
Physical Exam
8
Differences between SDS and FIM

• Merges 12 different domains into 3 types (
  Demographics)
• Drops 2 character domain prefix adds as
  variables
• Patient population flags
• Precision variable
• Variable name changes due to generalization

9
Standards and the ChangeProcess
FDA
Industry
CDISC Standards
10
Issues

• FIM acceptance by Statisticians
• FIM Compatibility with current DM/Stats processes
  and sponsor standards to produce CRTs and
  Analysis Files
• Impact on CDISC SDS team and implementers
• Where to draw the line between current models and
  FIM?
• Comprehensiveness SDS modeled only most common
  80/20 data variables per domain will that
  suffice for all data?
• What about other unique variables? Analysis
  variables?
• Other domains?
• Timeframe/transition SAS datasets require tools
  to be in place to be useful (for viewing,
  merging, etc.)
• Would FDA accept CDISC SDS or FIM format in
  interim?

11
Options for Moving Forward

• Freeze CDISC SDS at V 2.0 cease model
  development until FDA publishes guidance
• SDS can continue with implementation guidance
• Publish V 2.1 as CDISC Doc only
• FDA Publishes FIM
• FDA accepts V 2.1 as well as FIM format files?
• Publish V 2.1 as CDISC/HL7 Doc
• Publish V 3.0 based on FIM as HL7 Metadata
  Document
• RCRIM Publishes FIM