The Semantic Foundation to

Interoperable Clinical and Research

Information Systems in Cardiology
Brian McCourt1, James Tcheng, MD2, David Kong,
MD1, Bron Kisler3
1Duke Clinical Research Institute, 2Duke Translational Medicine
Institute, 3Clinical Data Interchange Standards Consortium
Robert M. Califf, Robert A. Harrington, Leanne K. Madre, Eric D. Peterson, Deborah
Roth and Kevin A. Schulman. Curbing The Cardiovascular Disease Epidemic: Aligning
Industry, Government, Payers, And Academics. Health Affairs, 26, no.1 (2007):62-74.
htt // t th lth ff i / i/ i t/26/1/62
Collect Once, Repurpose Many Times

Clinical
Decision
Quality Support Public &
Measurement Population
& Patient Health
Safety

Business,
Operations/
Research
Administration

Clinical Data
Don Mon, AHIMA
Groundbreaking Advances

“The adoption of standards and common

data elements across diseases is
groundbreaking, promotes cross-disease
analysis, and provides a rich source of
information to be mined by researchers
around the world.”
Barbara M. Alving, M.D., Acting Director, NCRR
NIH Launches Clinical Studies Nationwide to Investigate Rare Diseases
$71 Million Effort to Address Neglected Conditions (Friday, May 5, 2006)
President’s Council of Advisors on Science and
Technology (Dec 2010)

„ Atomic data elements as the unit of exchange

„ 2 use cases related to these projects
CV Standards Background

„ Long history of data standards work by Duke and

cardiovascular professional societies, lead by ACC.
„ 2006-2008: Under NIH Roadmap contract, Duke
partnered with CDISC and HL7 to develop data
standards for Tuberculosis & Cardiology (ACS
specifically)
z Focus on methodology. 1st time for clinical content
as “Data Standard”
z May 2008 approved as joint CDISC (Clinical
Research) & HL7 (Healthcare) standards
„ 2008-2010: Multiple emerging initiatives
„ 2010: Opportunity to expand the content of these
standards and align to meet needs of broader
stakeholder community
Mission

Develop and maintain a consensus set

of cardiology data standards as a
single collaborative set of work
products serving stakeholders across
the healthcare industry
Topics

„ Cardiology Data Standards Program

Broad multi-stakeholder project to develop
cardiology data standards across healthcare

z Project stakeholders, objectives and

deliverables
z Cardiology Domain Analysis Model, R2
z HL7 & the healthcare pathway
z CDISC & the research pathway
 Completed
Project Overview In progress
In project scope
Cardiovascular Data Future

Non-specialty data Common cardiovascular clinical observations - Sub-specialty domains

CTN ACC/AHA NCRI Grant*
CDISC FDA Cardiac Imaging
BP /STS ACC/AHA/STS Registries
Future

Cardioverter defib procedures

18 total domains and growing

Exercise Electrocardiography
Congenital Heart Conditions
Top 100 EHR data elements

Corotid Artery Stenting and

Womens’ Presentation Sx
ACS History & Symptoms
Concomitant Medications

Endarterectomy (CARE)
Cardiac Cath and PCI

Echocardiography
STEMI/NSTEMI
Adverse Events
Demographics

(ICD Registry)
CV Outcomes

Coronary CTA

SPECT MPI
Vital Signs

ACC

(ACTION)

(IMPACT)
(CathPCI
Stroke
CDISC
National Cardiovascular
CDASH Standards
Data Registries

*National Cardiovascular Research Infrastructure

CV Clinical CDISC HL7

- Data Elements - SDTM standard for FDA submission - Mappings to HL7 RIM & EHR stds
- Event definitions - Controlled Terminology alignment - Adoption support for EHR’s
- Clinical terminology and - CRF templates - CCHIT EHR Certification (future)
data definitions - Stds adoption by researchers
Collaborating Organizations
„ American College of Cardiology Foundation (ACC)
„ Clinical Data Standards Interchange Consortium
(CDISC)
„ Duke Clinical Research Institute (DCRI)
„ Duke Translational Medicine Institute (DTMI)
„ US Food and Drug Administration (FDA)
„ Health Level Seven (HL7)
„ National Cancer Institute (NCI)
„ National Cardiovascular Research Infrastructure (NCRI)

Partial funding provided by NHLBI Award 1RC2HL101512-01

ACC: Data, Standards and Informatics

„ Journal of the American College of Cardiology

z http://content.onlinejacc.org/cgi/collection/data_standards
CDISC Standards & Therapeutic Area Initiatives

„ Clinical Data Acquisition

Standards Harmonization
(CDASH)
z Widely adopted data
elements for common
trial domains
z Actively developing
efficacy domains
„ Study Data Tabulation
Model (SDTM)
z Data model for
submission of trial
data to FDA
 Karen Hicks, MD. Standardized Data Collection For Clinical Trials. March 26, 2010
https://www.trialstransformation.org/standardized-data-collection-for-cardiovascular-trials-march-26-2010-meeting
FDA Cardiovascular Endpoints

„ Definition of Cardiovascular Death

„ Definition of Non-Cardiovascular Death
„ Definition of Myocardial Infarction
„ Common Classification Schemes for Myocardial Infarction
Categories
„ Definition of Hospitalization for Unstable Angina
„ Definition of Transient Ischemic Attack and Stroke
„ Definition of Heart Failure Requiring Hospitalization
„ Interventional Cardiology Definitions
„ Definition of Peripheral Arterial Revascularization Procedure
„ Definition of Stent Thrombosis
„ Bleeding Definitions
Work Products

Healthcare Research
Use cases and patient scenarios

Data elements and clinical definitions

Cardiovascular Domain Analysis
Model
Activity diagrams

Information model

Healthcare (HL7 RIM) representation HL7

Sample data collection forms CDISC

CDISC (CDASH, SDTM)
representations CDISC

Electronic metadata library HL7 and CDISC

publication
via NCI EVS/caDSR
Domain Analysis Model

„ Produced by subject matter experts

„ Represents the domain of cardiology using our
language (not technical speak)
„ Explicitly describes requirements
„ Is the medium that requirements are communicated
to technical experts
Scope & Use Cases

„ Scope
z General cardiovascular care (broad)

z Ischemic heart disease orientation (deep)

„ Use Cases
z Documentation of cardiovascular patient of
interest across providers
z Reporting to a quality improvement registry

z Reporting to a clinical trial

z Submission of clinical trial with cardiovascular

endpoints to the FDA
Use Cases & Storyboards

„ Role
z Examples of typical real life scenarios

z Text descriptions of the work processes that an

information system needs to fit into
„ Product
z Written use case and storyboard documents
Use Cases & Storyboards
Data Elements

Data Element Name: History of peripheral vascular disease

Clinical Definition: Indicate if the patient has a history of peripheral vascular disease. This
can include:

1. Claudication either with exertion or at rest.

2. Amputation for arterial vascular insufficiency.
3. Aorto-iliac occlusive disease reconstruction, peripheral vascular bypass surgery,
angioplasty or stent; or percutaneous intervention to the extremities.
4. Documented abdominal aortic aneurysm (AAA) repair or stent.
5. Positive non-invasive/invasive test.

This does not include procedures such as vein stripping, carotid disease, or procedures
originating above the diaphragm.

Valid Values: Yes, No

Merging Killip Class across multiple databases

Variable Name Type / Values Description / Derivation

KILLIPB Numeric Baseline Killip Class

GUSTO IIb
1=Killip class I GUSTO2B.KILLIPB (3=I, 4=II, 5=III, 6=IV) **at randomization

2=Killip class II Otherwise missing

PURSUIT
3=Killip class III ****IMPUTED KILLIP CLASS****

4=Killip class IV QSRAND.RALES (3=None, 4=<= 1/3 up, 5=>1/3 up)

DEMO.DSYSBPB
Killip 1 = RALES=3 AND DSYSBPB>=80
Killip 2 = RALES=4 AND DSYSBPB>=80
Killip 3 = RALES=5 AND DSYSBPB>=80
Killip 4 = DSYSBPB<80
**rales and SBP at randomization
PARAGON A
BASELINE.KILLIPB (1=I, 2=II, 3=III, 4=IV) **at randomization
PARAGON B
DEMOG.KILLIP (1=I, 2=II, 3=III, 4=IV) **at time of index event.
GUSTO IV ACS
N/A
SYNERGY
HISTBASE.KILLIP * at time of randomization

Example courtesy of Karen Pieper

Aggregation & Harmonization Process
NCRI Data Standards
1. EHR Workgroup
Top 100 Definitions are
recommended by expert
2. Select NCDR panel and formally vetted
via CDISC & HL7 public
elements for
processes to produce
harmonization consensus version. ACC
and inclusion in serves as stewarding
standard organization of resulting
standards.
3. Elements
needed to
support FDA
Endpoint
definitions 5. Maintenance

4. CDISC & HL7 Standards

Development Processes
Harmonization is hard work!

„ Many sources, inconsistent definitions. Often for a

good reason.
Sample Data Collection Forms

„ Case Report Form (CRF) mockup

z CRF’s are familiar to many stakeholders

„ Highlights metadata and relationship issues across

domains
„ Highlights requirement to align data elements and
clinical workflows
Data Elements
Activity Diagram

„ Role
z Represent patient care processes

z Recognize the data in context of the workflow

„ Product
z CV Activity diagrams based on care guidelines
and common practices
Activity Diagram
Information Model

„ Role
z Represent scope of the data used in the domain

z Represent the relationship among data elements

z Mechanism to load the content into a metadata

repository (EVS/caDSR, CDISC SHARE)

„ Product
z Information model for cardiovascular disease including
diagram, classes, attributes, descriptions
Information Model

Cardiology (Acute Coronary Syndrome) Domain Analysis Model, Release 1

Information Model: Design Notes

„ Using Enterprise Architect as UML modeling tool

„ Includes annotations of NCI EVS Vocab, HL7 Reference
Information Model, CDISC SDTM and use cases
„ Following NCI modeling conventions to support the metadata
publication process
z Semantic Integration Workbench (SIW) process for
loading model into the Enterprise Vocabulary
Server/cancer Data Standards Repository (EVS/caDSR)
„ Add-on tooling: build a back-end EA reporting function to
generate clinician friendly data element reports (to replace
Excel data element curation spreadsheets)
NCI EVS\caDSR Data Element Representation
HL7

„ Clinical Interoperability Council

z Forum and portal for domain experts to
contribute clinical knowledge to using HL7
Development Framework (HDF)
z Drafting a DAM Modeling Guide

z Contributing requirements, and typically

information models, to NCI’s EVS/caDSR
„ Detailed Clinical Models
z Highly specific, free of context
Schedule

„ Spring/Summer: CDISC Public Comment

z Cardiology CDASH domains with SDTM
mapping
„ August 1 – September 5: HL7 Ballot Open
z Cardiology DAM, R2

„ October: Terminology/Metadata Publication

z Via UML semantic integration
Healthcare & Research Pathways
Future

Cardiovascular EHR
CCHIT Certification
Functional Profile
HL7 RIM
Representation RIM-derived
messages

NCI EVS/caDSR
CV Domain Terminology/
Analysis Model, R2 Metadata
Publication

FDA Cardiovascular
Trials Data
CDISC CDASH Warehouse
Representation SDTM Mappings
BRIDG Model
Thank You!

Brian McCourt
brian.mccourt@duke.edu