Clinical Trial Management System Working Group Meeting City of Hope National Medical Center

1
Clinical Trial Management SystemWorking Group
MeetingCity of Hope National Medical Center
Joyce C. Niland, PhD Chair, Information Sciences
February 19, 2004
2
Development Team

• Dr. Joyce Niland
• Dr. Doug Stahl
• Dr. David Ikle
• Dr. Hemant Shah
• Cindy Stahl
• Joycelynne Palmer
• Amy Cox
• Stacy Berger
• Tremon White
• Tuong Tieu

• Chair, Information Sciences
• Director, Biomedical Informatics
• Associate Director, Biostatistics
• Information Scientist
• Systems Analyst
• Systems Analyst
• Systems Analyst
• Database Architect
• Application Developer
• Application Developer

3
Outline of Todays Talk

• Past Experience in Clinical ResearchSystem
  Development
• Current Approach Fully Integrated Research
  Standards Technology (FIRST)
• Proposed Future Approach Year One

4
City of Hopes Experience inClinical Trials
System Development
5
Research Protocol Lifecycle
Results
Research Idea
lt---Protocol Data---gt
lt----Patient Data----gt
6
City of Hope Experience in Clinical Research
System Development and Deployment

• Computerized System Year Deployed
  
• Biostatistics Information Tracking System
  (BITS) 1989
• Frozen Tumor Bank Specimen Tracking System
  1992
• NCCN Web-based Outcomes Research System
  1996
• Forms Tracking System 1997
• Clinical Trials On-Line Protocol Delivery System
  1998
• Clinical Pathways Patient Management System
  1999
• Web-enabled Genotype-Phenotype Database
  2000
• Guideline Graphical Decision Support Interface
  2001
• GCRC Protocol Patient Administrative System
  2002
• Web-based Services for Protocol Data Integration
  2002
• Metadata Repository System
  2003
• Prototype

7
Clinical Trials System at COH

• Over 400 ongoing trials at City of Hope
• 30-40 of COH patients enroll on one or more
  trials
• 1/3 pharma, 1/3 cooperative group, 1/3 intramural
• Data Coordinating Center for 60 multi-center
  trials
• Created centralized clinical trials database
• Biostatistics Information Tracking System (BITS)
• In continual use for all COH trials since 1989
• Over 500,000 records on more than 20,000 patients
• Migrated to MS SQL server, building web
  interfaces
• Also deploying scannable forms

8
BITS Clinical Trials On-Line support gt400
clinical trials gt110 investigators
9
Features of BITS

• Protocol accessions, demographics, labs, and
  relapse survival on all patients
• Detailed case report form (CRF) data for patients
  on in-house trials
• Export to NCI CDUS application
• Mirrored copy of Oacis to import
• A2K demographics
• Sunquest labs
• Cytogenetics
• HLA data

10
Deployed for all Pediatric protocol patients
11
Lessons Learned....

• Architecture must support within trial and
  cross-protocol analyses

Study-Specific Model for
Unified Data Model for
T R I A L 1
T R I A L 2
T R I A L 3
T R I A L 4
T R I A L 5
T R I A L 6
Case Series, outcomes research, meta analysis,
data mining
Case series, outcomes research, meta analysis,
data mining
12
Lessons Learned.

• Metadata are critical!
• Data about the data themselves
• Technical directory system perspective
• Business directory user perspective
• Required for critical database functionality
• Merging in legacy electronic data
• Data directory to facilitate data mining
• Management of data derivation and logic rules
• Preparing summary versions of massive data sets
  (data marts)
• Documenting creation and sunset dates of data
  elements codes

13
Creation of an Effective Metadata Repository

• Building an interactive Web-based metadata
  repository (ISO/IEC 11179 compliant)
• Will be the driver of future systems
• Technical metadata populated from feeder
  systems
• Detailed analysis of business metadata of all
  fields underway by full-time Metadata Analyst

14
Root Cause of Data Managementand Integration
Problems

• Lack of a unified information architecture
• Requires standard
• Data model
• Vocabularies
• Data interfaces
• Improved information technology infrastructure
  and toolsneeded to exploit significant
  investments in biomedical data
• Formidable task
• Becomes more intractable and costly as time goes
  on

15
FIRSTFully Integrated ResearchStandards
Technology
16
FIRST

• 3-year GCRC supplement grant funded by the NCRR
• Goal Develop a fully integrated information
  management and decision support environment
• Facilitate all phases of clinical research
• Support research infrastructure of the GCRCs
• Development group COH Division of Information
  Sciences
• Input and testing from USC and CHLA
• Collaboration with Dr. David Forslund, Advanced
  Computing Lab, Los Alamos National Laboratory

17
Objectives of FIRST

• Create a clinical research environment to support
  the wide range of clinical research activities
  across the protocol life cycle
• Bring together enabling technologies
• Standardized information modeling
• Object-oriented distributed architecture
• Advanced graphical user interfaces
• Ultimate goal scalable interoperable system

18
caBIG Desiderata

• Open access, open source
• Derived from common information models
• Standards for data exchange formats
• Data and metadata following ISO/IEC 11179
• Consume appropriate public, open access standards
  when available
• All of these principles are being followed in
  FIRST

19
Through FIRST

• A common standardized UML model for clinical
  research will be made publicly available
• Semantic analysis will yield common standards and
  mapping among vocabularies
• Integration of existing data sources will be
  facilitated
• Prototype applications that support the clinical
  research life cycle are being developed and tested

20
Functionality to be Supported in FIRST
Results
Research Idea
21
Principles in Developing FIRST

• Computational and semantic standards
• Information integration
• System interoperability
• Workgroup collaboration
• Sufficiently flexible methods to
• Scale over time
• Adapt to unanticipated needs
• Accommodate multiple disease entities
• Adopt to different research environments
• Exploit major advances in science and technology

22
FIRST Unifying Information Architecture

• Enforce both technological and lexical standards
  through
• Standardized Data Model
• Semantic Standards
• Technical Interoperability

23
FIRST Schematic Diagram
24
FIRST Modeling Process
25
Standardized Data Model

• Requires clear understanding of the environment
• FIRST Advisory Group of senior investigators
• Clinical trialists, biostatisticians, research
  staff, informaticists
• Originally envisioned units of application
  functionality as shown in protocol life cycle
• Actual components of model far less orderly,
  organized, and separated in practice
• Much time and effort required to
• Separate, document, and specify units of
  application functionality
• Define interrelationships sufficiently begin
  application development

26
30,000 Foot View of FIRST Information Model
ProtocolRegistered
Available for Accrual
Available for Accrual
ProtocolStatus
SAE onProtocol
Conditions met,N slots available
Registration
ReportedSAE
N 1Slots Available
Suggestedfor Protocol
First Study ParticipantStatus
FullyEligible
Consented
Registered
RXHeld
On-Study
. . .
Treatment
SAE
Eligibility Screening
ConsentingProcesses
AccessioningProcesses
Eligibility Filtering
SAEForm
Data CollectionProcess
ConsentForm
Eligibility Checklist
Form A
27
FIRST Modeling Process

• Requirements analysis and documentation
  conductedin a uniform manner
• Unified Modeling Language (UML)-based clinical
  research domain model
• Maintains consistency among model, metadata,and
  technology
• Assures architectural soundness
• Improves communication between domain technical
  experts
• Facilitates translation into an executable
  program
• Promotes platform independence
• City of Hope members of HL7 working group
• Expert consultant Gunther Schadow, Regenstrief
  Institute

28

• Eligibility Filtering Use Case
• Use Case 20.0 Protocol Abstraction for
  Eligibility Filtering
• Description
• The protocol abstractor entered data about the
  protocols eligibility and exclusion criteria into
  the system.
• Actor
• Protocol Abstractor
• System
• Pre-Conditions
• Protocol is registered into the system.
• Basic Course
• .The actor has logged into the system (See Use
  Case 1.0), searched and selected the desired
  protocol (See Use Case10.0) (example screen 3)
• .The system displays a page (example screen 4)
  that allows the actor to select Eligibility
  Criteria within the Protocol Abstraction
  function.
• .The system displays the protocol abstraction
  eligibility filtering criteria page (example
  screen 22, 27, 23, 32). The following
  information is requested about the protocols
  eligibility and exclusion criteria. All criteria
  are required. NA is used if the protocol does
  not specify.
• Gender (Alternate Course 8.0.3 Add a value to
  any of the pick lists)
• Minimum age and unit
• Maximum age and unit
• Functional status
• Minimum life expectancy

29
HL7 RIM 2.01 Class Diagram
30
HL7 Reference Information Model(RIM 2.01)
Backbone
31
HL7 Block Diagram
32
FIRST Block DiagramProtocol Registration
33
FIRST Semantic Standards
34
No Single VocabularyCan Meet All Needs

• More than one vocabulary system required for
  FIRST
• SNOMED, LOINC, RX-NORM
• caDSR Common Data Elements (CDEs)
• External standards can be represented as CDEs
  (e.g. ICD-O-3)
• Still may need to additional terms/concepts
• Utilize UMLS to search for terms

35
UMLS for Mapping AcrossTerms and Concepts
UMLS Metathesaurus
Vocab. Source A
Concept 1
Term 1
Concept 2
Term 2
Vocab. Source B
Concept 3
Term 3
Concept 4
Term 4
36
Without proper metadata
37
Sample Master Data Elements
Source HL7 Ref
38
FIRST Technologies
39
FIRST Technologies

• Object Management Group (OMG) International
  Standards Organization
• Develops technically integrated, commercially
  viable, vendor independent specifications for
  software industry
• Achieved international consensus on Common Object
  Request Broker Architecture (CORBA)
• Facilitates technical interoperability among
  disparate information systems
• Recognized as a potential solution in healthcare
  informatics
• Has not been operationalized in clinical research
  context

40
FIRST Technologies

• Software architecture based
  on international standards
  interoperable components
• Leveraging existing standardsor help expand/
  build new ones
• Systems are future proof if retain
  CORBA-compliant interfaces

41
OpenEMed (formerly TeleMed)

• An intuitive patient-record system that supports,
  image, audio, and graphical data
• Open-source, interoperable components based on
  OMG Healthcare Domain Taskforce interface
  standards
• Enables multiple databases to be integrated
  together to create a virtual patient record
• Individual healthcare facilities still own and
  manage their own data
• Data accessible to others who have treated the
  same patient
• Security of the data is maintained, patient
  privacy and confidentiality is ensured

42
OpenEMed (formerly TeleMed)

• CORBA Services used in FIRST
• PIDS Person Identification Service
• Correlate health records among multiple
  institutionbased on a "core" set of profile
  elements, whileprotecting confidentiality
• COAS Clinical Observation Access Service
• Interface to supply clinical observations,
  includingraw data recordings, and derived
  judgmentsor knowledge

43
Applying OpenEMed to FIRST Global Model
PIDS
ProtocolRegistered
Available for Accrual
ProtocolStatus
Available for Accrual
SAE onProtocol
COAS
Conditions met,N slots available
Registration
ReportedSAE
N 1Slots Available
Suggestedfor Protocol
PIDS
First Study ParticipantStatus
FullyEligible
Consented
Registered
RXHeld
On-Study
. . .
Eligibility Screening
ConsentingProcesses
AccessioningProcesses
Treatment
SAE
COAS
Eligibility Filtering
COAS
SAEForm
Data CollectionProcess
ConsentForm
Eligibility Checklist
Form A
44
Significance of FIRST Technologies

• Technology behind FIRST application
  prototypesmuch more significant than their
  applied functionality
• Significant advantages through flexibility and
  re-usability of CORBA services
• Minimizes additional development later (e.g.
  security and auditing capability)
• Places emphasis on knowledge representation,
  rather than database construction

45
Graphical Representation of PIDS
Defines relationships among data elements
Tells service what the data is

• Data
• patient

PIDS DB
PIDS (service)
Entity-Attribute-Value (EAV) structural
independence Stores data preserves
relationships defined by the configuration file,
maintained by PIDS service
Can be HL-7, flat files, etc
46
Protocol Abstraction UnderliesAll FIRST
Applications

• Abstraction Categories
• Administrative Overview
• Study Personnel
• Protocol Entities / Organizations
• Study Design
• Scientific Objectives
• Eligibility Criteria
• Statistical Approach
• Treatment Roadmap
• Toxicity Monitoring
• Outcomes Assessment
• Data Collection Expectation
• Obvious logical extension protocol authoring
  via FIRST

47
Sample FIRST Application
48
Example Screen 22 Protocol Abstraction -
Eligibility Filtering Criteria pg 1
Protocol Participant Reports
Other Main Menu Exit
Overview Information Related Entities, Committees
Organizations Protocol Personnel Scientific
Abstraction Eligibility Criteria Statistical
Approach
Protocol Registration Protocol Abstraction Protoco
l Management Tasks
Eligibility Filtering Criteria Oncologic
Studies Demographics
Minimum
Maximum
3
18
Gender
Age
note If physiologic age is used, enter the
upper limit 5 years for the maximum
Status
Check all eligible functional status
descriptions
Able to carry on normal activity and to work no
special care needed
X
Unable to work able to live at home and care for
most personal needs varying amount of assistance
needed
X
Unable to care for self requires equivalent of
institutional or hospital care disease may be
progressing rapidly

3
Minimum life expectancy
And above
Click here to communicate with the System
Administrator
49
Example Screen 27 - Edit/View Protocol
Abstraction - Participant Characteristics for
Eligibility Filtering - Disease List
Protocol Participant Reports
Other Main Menu Exit
Overview Information Committees
Organizations Protocol Personnel Scientific
Abstraction Eligibility Criteria Statistical
Approach
Protocol Registration Protocol Abstraction Protoco
l Management Tasks
FIRSTSM Unique Protocol ID
Prefill from search
This protocol is available for accrual
NCI Disease List Check all eligible
Any Hem
Any Diagnosis
Any Solid
X
Hematologic
Solid Tumors
Esophagus
Prostate
Hematopoietic, other
Anus
Eye and Orbit
Rectum
Leukemia, Lymphoid
Respiratory intrathoracic Organs, other
Bones Joint
Genital - Female, other
Leukemia, Monocytic
Brain Nervous System
Genital - Male, other
Skin Melanoma
Leukemia, Myeloid
Breast - Female
Ill-defined Sites
Skin, other
Leukemia, not otherwise specified
Breast - Male
Kaposis Sarcoma
Small Intestine
Lymphoma, Hodgkins
Buccal Cavity Pharynx
Kidney
Soft Tissue
Lymphoma, Non-Hodgkin's
Larynx
Cervix
Stomach
Mycosis Fungoides
Thyroid
Colon
Liver
Multiple Myeloma
Unknown Sites
Corpus
Lung
Urinary Bladder
Digestive Organ, other
Ovary
Endocrine System, other
Urinary, other
Pancreas
Click here to communicate with the System
Administrator
50
Example Screen 50 - Protocol Filtering Search
Results
Protocol Participant Reports
Other Main Menu Exit
Protocol Matches for
FIRSTSM Unique Person ID 46586
Name Kelly M Clarkson
Status
Gender
Age
Evaluable
Min Life Expect
Diagnosis
Measurable
Metastasis
F
20
No Assist
6 mo
Lymphoma
Yes
Yes
Pt ID(Prefill pt ID )
Protocol Information
N/A
to 18yrs
No Assist Some Assist
Lymphoma
N/A
Yes
Yes
3 mo
113
N/A
18yrs to 70yrs
No Assist Some Assist
All Hem
N/A
N/A
N/A
3 mo
540
N/A
18yrs to 65yrs
No Assist Some Assist
All Hem
N/A
N/A
N/A
N/A
700
N/A
18yrs to
No Assist Some Assist
Any Cancer
N/A
N/A
N/A
N/A
003
Filter Summary 15 protocols in FIRST10
available for searching 5 matched your query
Click here to communicate with the System
Administrator
51
(No Transcript)
52
(No Transcript)
53
(No Transcript)
54
(No Transcript)
55
(No Transcript)
56
Possible caBIG DevelopmentApproach Year One
57
Proposed Approach to caBIG
58
Possible Iterative Approach to caBIG Development
Adopters
City of Hope
Working Group
Univ of Pitt
NCI
59
Discussion
60
FIRST System Design

• 3-tier architecture

Presentation Layer Java-based
Middle TierXML Process DefinitionLanguage
Rules Engine
Data Layer CORBA / OpenEMed