Anita WaldenDuke Translational Medicine

Institute

TuberculosisData Standardization and

Harmonization

Supported by BAA-RM-04-23

“Assessing Critical Research Needs in TB Diagnostics in HIV-

Infected and Uninfected Children”

Kirkland Center, National Labor CollegeSilver Spring, MD

June 30, 2011

Background & Project Objectives

• Funded by NIH Roadmap contract – Carol Dukes Hamilton PI

• Focus on methodology for developing therapeutic area data standards

• Produce real, useful products along the way• Report on experience, process, best practices

back to the larger community

Purpose:

Patient

Clinician

Healthcare Data Systems

Patient care

Quality Improvement

Research

Reimbursement

Post Marketing Safety

Decision Support

Administration & Mgt.

Public Health Reporting

…

Data Uses

Single Source Multiple Uses

SHELL…

EmitterReceiver

…shall we go now ?

Semantic Interoperability(the Human Factor)

?

EmitterReceiver Term

DefinitionConcept

Semantic Interoperability(in eClinical Systems)

• Unique Preferred Term (PT)

• Consensus-based Definition

• Unique Numeric Code (CUI)

• Associated Synonyms

• Accessible Publishing Env.

AE Relatedness Example

Company 2• Not Related• Doubtful• Possible• Very Likely• Probable

Company 3• NO• YES / Unknown

AE Relatedness to Study Drug

(no standard codelist defined)Company 1• No• Unlikely• Possible• Definite• Probably

AE Relatedness Example

Standard Codelist• No• Unlikely• Possible• Definite• Probably

AE Relatedness to Study Drug

(with standardized codelist)

Data element: chest x-ray result

Funding &Development

PartnersTB ResearchSurveillance

Patient Care Regulatory

Data element: chest x-ray result

Clinical Trial #3Clinical Trial #2Clinical Trial #1

Clinical Trial #4 Clinical Trial #5

= = = =

Data Mapping

Data element: chest x-ray result

Funding &Development

PartnersTB ResearchSurveillance

Patient Care Regulatory

Global TB Data Standards:an agreed upon set of common data elements,

how they are defined, how they “look” electronically

Patient Care

World

Clinical Research

World

• Healthcare data from multiple data sources• Data is patient-centric• HL7 Pervasive Standard• Medical Records assembled• from multiple Sources

• Research data is carefully controlled• Data is trial-centric• CDISC/CHI Emerging Standards• Data flows from site to CRO to

sponsors to regulatory

Connecting Parallel Universes

Needed…shared semantics

…common data elements

Goals

• Focus on methodology for developing therapeutic area data standards

• Produce useful products for TB• Report experience, process, best practices

Project StepsCreated stakeholder groupCreated master data element list from healthcare and related secondary usesDefine a set of data elements according to the ISO 11179 standard (TB139 data elements)Developed consensus clinical definitions for the data elementsCreated a dynamic model of the domain represented through an activitydiagram, using the UML standardCreated a research representation of the data elements in the CDISC SDTM model.Made the resulting standards available to the public through an ANSI accredited balloting processStorage in NCI Meta-data Repository

Methods

• Standardize at source healthcare– Data element as unit of exchange– Specificity sufficient for semantic interoperability– Work within HL7

• Include all stakeholders– Research representation CDISC– Public Health Representation CDC– Quality Imp. Professional Societies

Data Element Contributors– World Health Organization

Stop TB Partnership• DOT• DOT Plus

– Center for Disease Control and Prevention (CDC) • TIMS – Meta data• NEDSS – Meta data• RVCT - forms

– MOXI Brazil CRF forms– TB Alliance

• TBESC - forms• TBTC Study 26 – Meta data• TBTC Study 27 – forms

Expert GroupRepresented Organizations

• Aereas Global TB Vaccine Foundation

• Center for Disease Control and Prevention (CDC)

• Clinical Data Interchange Standards Consortium (CDISC)

• Clinical research experts• Duke University Medical

Center • Foundation for Innovative

New Diagnostics (FIND)

• Global Alliance for TB Drug Development

• KNCV• Health Level 7 (HL7)• National Cancer Institute• National Heart, Lung & Blood

Institute (NHLBI)• National Institutes of Health

(NIH)• National TB Controllers

Association• WHO’s Stop TB Partnership• Pharmaceutical Companies

Artifacts for Standards

Use cases and story scenarios (H)

Data elements and clinical definitions (U)

Domain class model (H)

Domain activity diagram (H)

Data collection forms (R)

Suggestions for SDTM representation (R)

H = Healthcare / R = Research / U = Universal

Example Model Element Description

• Person.birthDate– Definition: the calendar date corresponding to when the

person was born.– Examples: 06/11/1954; 1976; 12/1957– Comment: birthdate may be a partial date when only the

calendar year or calendar month and year are known.– Constraints: birthdate shall not exceed current date.

birthdate must be less than or equal to person.deceaseDate

– Rationale: birth date is needed to approximate a person’s age for use in clinical decision making

99 Total Doses of Tuberculosis Treatment

Total number of all Tuberculosis doses, typically derived from medication logs, to indicated if full doses of all Tuberculosis drugs has been taken.

numeric field

100 Tuberculosis Treatment Status

The state or condition of Tuberculosis drug treatment for active Tuberculosis, denoting ongoing, completed or unknown status

Treatment completed; Treatment interrupted; Treatment discontinued; Treatment failure; Cure; Transferred out; Death; Absconded/Defaulted; Clinical Cure

101 Did the Subject Complete Treatment for Tuberculosis?

Subject received all anti-Tuberculosis medication, as prescribed

No; Yes; NA; Unknown

102 Tuberculosis Treatment Failure Reason

Specified reason assigned to explain microbiological or clinical failure of Tuberculosis drug therapy to have resulted in culture-conversion to negative with 16 weeks of Tuberculosis drug initiation, or clinical improvement within the same time frame.

Primary Drug resistance; Secondary Drug resistance; Inadequate serum drug levels; Non-compliance; Extensive disease; unknown; Other (specify)

103 Reason for Study Tuberculosis Treatment Discontinuation

A statement offered in explanation or justification for permanent discontinuation of one or more Tuberculosis drugs being given in the context of a research study

Refused study therapy; Pregnant; Death; Withdrew consent; Physician stopped study Rx/MD Decision; Lost to follow up; Adverse effect; Subject moved; Incarceration; False-positive culture

Data Elements

TB Class Model

TB ActivityModel

Rationale: show clinical processes of the Domain

Research Representation

Rationale: design with the end in mindProduct: 1.) data collection form

2.) SDTM suggestions

TB in SDTM, like this?

Benefits

• Facilitate Collaboration Between Organizations• Promote Secondary Analysis (Aggregate data from multiple

studies)

• Use to Build Repositories for Data and Samples• International Accessibility • Facilitate the Drug Submission and Approval

Process• Reduce Costs

Resources

• Critical Path (NIH,FDA) • CDISC – Starting Therapeutic Standards for

Research• HL7 – Clinical Interoperability Council – Ballot as

National or International Standard for Healthcare• NCI – Storage of Data Elements for Public

Accessibility and Generation of Tools.• Duke Translational Institute

Information

• HL7 Website (http://www.hl7.org/implement/standards/v3messages.cfm)

• Anita Walden PM for TB Data Standard ProjectCo-Chair HL7 Clinical Interoperability Council Informaticists, Duke Translational Medicine Institute

anita.walden@duke.edu919.672.6587